inst/shiny-examples/main/app.R
In nickreich/clusterPower: Power Calculations for Cluster-Randomized and Cluster-Randomized Crossover Trials
#
# This is a Shiny web application. You can run the application by clicking
# the 'Run App' button above.
#
# Find out more about building applications with Shiny here:
#
# http://shiny.rstudio.com/
#
library(clusterPower)
library(shinythemes)
library(shiny)
library(shinyBS)
library(promises)
library(future)
library(future.callr)
library(shinyjs)
library(V8)
library(ggplot2)
plan(callr)
# labels for arguments
simnsubjectstext <- "Number of observations per cluster (nsubjects)"
refsimnsubjectstext <-
"Number of reference arm observations (nsubjects)"
treatsimnsubjectstext <-
"Number of treatment arm observations (nsubjects)"
simnclusterstext <- "Number of clusters per arm (nclusters)"
refsimnclusterstext <-
"Number of reference arm clusters (nclusters)"
treatsimnclusterstext <-
"Number of treatment arm clusters (nclusters)"
analyticnsubjectstext <- "Mean observations per cluster (nsubjects)"
analyticnclusterstext <- "Mean clusters per arm (nclusters)"
analyticICCtext <- "Intracluster correlation coefficient (ICC)"
analyticsigma_sqtext <- "Within-cluster variance (sigma_sq)"
analyticsigma_b_sqtext <- "Between-cluster variance (sigma_b_sq)"
treatsigma_sqtext <-
"Treatment arm within-cluster variance (sigma_sq)"
refsigma_b_sqtext <-
"Reference arm between-cluster variance (sigma_b_sq)"
treatsigma_b_sqtext <-
"Treatment arm between-cluster variance (sigma_b_sq)"
simnsimtext <- "Number of simulations (nsim)"
refmutext <- "Reference arm expected mean (mu)"
treatmutext <- "Treatment arm expected mean (mu2)"
refICCtext <- "Reference arm ICC (ICC)"
treatICCtext <- "Treatment arm ICC (ICC)"
refp1text <- "Reference arm estimated proportion (p1)"
treatp2text <- "Treatment arm estimated proportion (p2)"
refc1text <- "Reference arm expected count (c1)"
treatc2text <- "Treatment arm expected count (c2)"
narmstext <- "Number of arms (narms)"
delimtext <- "Note: comma delimited"
dtext <- "Expected difference in arm means (d)"
didsigma_b_sqposttext <-
"Post-treatment between-cluster variance (sigma_b_sq1)"
didsigma_b_sqpretext <-
"Pre-treatment between-cluster variance (sigma_b_sq0)"
ntimestext <- "Number of measurement time points (ntimes)"
varttext <- "Total variation of the outcome (vart)"
rho_ctext <-
"Baseline and post-test cluster-level correlation (rho_c)"
rho_stext <-
"Baseline and post-test subject-level correlation (rho_s)"
stepstext <- "Number of crossover steps (steps)"
mu0text <- "Estimated baseline mean (mu0)"
mu1text <- "Estimated post-treatment effect (mu1)"
p0text <- "Estimated baseline proportion (p0)"
p1text <- "Estimated post-treatment effect (p1)"
c0text <- "Estimated baseline proportion (c0)"
c1text <- "Estimated post-treatment effect (c1)"
refsigma_sqtext <-
"Reference arm within-cluster variance (sigma_sq)"
nsubjectsirgttunclusttext <-
"Observations in unclustered arm (nsubjects)"
ncontrolsirgttunclusttext <-
"Observations in unclustered arm (ncontrols)"
nclustersswtext <- "Number of clusters (nclusters)"
nsubjectsdelim <-
"Comma delimited. Length of nsubjects can be 1 (all equal) or equal to the number of clusters."
ui <- fluidPage(
theme = shinytheme("united"),
h1(
id = "big-heading",
"Power Estimation for Randomized Controlled Trials: clusterPower"
),
tags$style(HTML("#big-heading{color: #337ab7;}")),
shinyjs::useShinyjs(),
sidebarLayout(
sidebarPanel(
tags$head(
tags$style(type = 'text/css',
'form.well { max-height: 700px; overflow-y: auto; }')
),
selectInput(
"type",
"CRT Type",
choices = c(
"Parallel",
"Multi-Arm",
"Difference-in-Difference",
"Stepped Wedge",
"Individually-Randomized Group"
)
),
selectInput(
"dist",
"Outcome Distribution",
choices = c("Normal", "Binary", "Count")
),
selectInput("meth", "Method",
choices = c("Analytic", "Simulation")),
# Below values can be reset to defaults using the restore defaults button
div(
id = "allValues",
shinyjs::hidden(numericInput("power", "power", value = NA)),
#input for cpa.normal -------------------------------------------------------------
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Normal' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpanormal", analyticnclusterstext, value = 10),
#nsubjects
numericInput("nsubjectscpanormal",
analyticnsubjectstext,
value = 20),
# CV
numericInput("CVcpanormal", "Coefficient of variation (CV)", value = 0),
bsTooltip(
"CVcpanormal",
"When CV equals 0, all clusters are the same size.",
'right',
options = list(container = "body")
),
# d
numericInput("dcpanormal", dtext, value = 0.43),
# ICC
numericInput(
"ICCcpanormal",
analyticICCtext,
value = NA,
step = 0.01,
min = 0,
max = 1
),
# variance params
numericInput(
"sigma_sqcpanormal",
analyticsigma_sqtext,
value = 0.01,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sqcpanormal",
analyticsigma_b_sqtext,
step = 0.001,
value = 0.1,
min = 0
),
numericInput("vartcpanormal", varttext, value = NA)
),
# input for cps.normal
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsnormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
tags$style(HTML("#armHead{color: #337ab7;}")),
# nclusters
numericInput("nclusters1cpsnormal", refsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects1cpsnormal",
refsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects1cpsnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu
numericInput("mucpsnormal", refmutext, value = 2.4),
# variance params
numericInput(
"ICCcpsnormal",
refICCtext,
value = NA,
step = 0.01,
min = 0,
max = 1
),
bsTooltip(
"ICCcpsnormal",
"Intracluster correlation coefficient",
'right',
options = list(container = "body")
),
numericInput(
"sigma_sqcpsnormal",
refsigma_sqtext,
value = 0.2,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sqcpsnormal",
refsigma_b_sqtext,
value = 0.5,
step = 0.001,
min = 0
),
### TREATMENT ARM
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsnormal", treatsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects2cpsnormal",
treatsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu
numericInput("mu2cpsnormal", treatmutext, value = 1.5),
# variance params
numericInput(
"ICC2cpsnormal",
treatICCtext,
value = NA,
step = 0.01,
min = 0,
max = 1
),
bsTooltip(
"ICC2cpsnormal",
"Intracluster correlation coefficient",
'right',
options = list(container = "body")
),
numericInput(
"sigma_sq2cpsnormal",
treatsigma_sqtext,
value = 0.2,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq2cpsnormal",
treatsigma_b_sqtext,
value = 0.5,
step = 0.001,
min = 0
)
),
# cpa.binary inputs start
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Binary' && input.meth == 'Analytic'",
# nuclusters
numericInput("nclusterscpabinary", analyticnclusterstext, value = 10),
# nsubjects
numericInput("nsubjectscpabinary",
analyticnsubjectstext,
value = 20),
# CV
numericInput("CVcpabinary", "Coefficient of variation (CV)", value = 0),
# ICC
numericInput(
"ICCcpabinary",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
),
# p1 & p2
numericInput(
"p1cpabinary",
refp1text,
value = 0.1,
step = 0.001,
min = 0,
max = 1
),
numericInput(
"p2cpabinary",
treatp2text,
value = 0.24,
step = 0.001,
min = 0,
max = 1
),
# other selections
checkboxInput("pooledcpabinary", "Pooled standard error", value = FALSE),
checkboxInput("p1inccpabinary", "p1 > p2", value = FALSE),
checkboxInput("tdistcpabinary", "Use t-distribution", value = FALSE)
),
# cps.binary inputs start
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsbinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpsbinary", refsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects1cpsbinary",
refsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects1cpsbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p1
numericInput(
"p1cpsbinary",
refp1text,
value = 0.8,
step = 0.001,
min = 0,
max = 1
),
# variance param
numericInput(
"sigma_b_sqcpsbinary",
refsigma_b_sqtext,
value = 1,
step = 0.001,
min = 0
),
## TREATMENT VALUES
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsbinary", treatsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects2cpsbinary",
treatsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p2
numericInput(
"p2cpsbinary",
treatp2text,
value = 0.5,
step = 0.001,
min = 0,
max = 1
),
#variance param
numericInput(
"sigma_b_sq2cpsbinary",
treatsigma_b_sqtext,
value = 1,
step = 0.001,
min = 0
)
),
# cpa.count input starts
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Count' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpacount", analyticnclusterstext, value = 10),
# nsubjects
numericInput("nsubjectscpacount",
analyticnsubjectstext,
value = 20),
# variance param
numericInput(
"CVBcpacount",
"Between-cluster coefficient of variation (CVB)",
value = 0.01
),
# expected outcome
numericInput("r1cpacount", "Reference arm mean event rate", value = 0.2),
numericInput("r2cpacount", "Treatment arm mean event rate", value = 0.35),
# other params
checkboxInput(
"r1inccpacount",
"Intervention probability < control probability",
value = FALSE
)
),
# cps.count input starts
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpscount",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpscount", refsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects1cpscount",
refsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects1cpscount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c1
numericInput(
"c1cpscount",
refc1text,
value = 200,
step = 1,
min = 0
),
#variance param
numericInput(
"sigma_b_sqcpscount",
refsigma_b_sqtext,
value = 0.5,
step = 0.001,
min = 0
),
## Treatment VALUES
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpscount", treatsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects2cpscount",
treatsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpscount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c2
numericInput(
"c2cpscount",
treatc2text,
value = 80,
step = 1,
min = 0
),
#variance param
numericInput(
"sigma_b_sq2cpscount",
treatsigma_b_sqtext,
value = 0.5,
step = 0.001,
min = 0
)
),
# cpa.ma.normal input starts
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Normal' && input.meth == 'Analytic'",
#narms
numericInput(
"narmscpamanormal",
narmstext,
value = 3,
max = 50,
min = 2,
step = 1
),
# nclusters
numericInput("nclusterscpamanormal", analyticnclusterstext, value = 10),
# nsubjects
numericInput("nsubjectscpamanormal",
analyticnsubjectstext,
value = 20),
# variance params
numericInput(
"varacpamanormal",
"Between-arm variance (vara)",
value = 0.02,
step = 0.001,
min = 0
),
numericInput(
"varccpamanormal",
"Between-cluster variance (varc)",
value = 0.01,
step = 0.001,
min = 0
),
numericInput(
"varecpamanormal",
"Within-cluster variance (vare)",
value = 0.1,
step = 0.001,
min = 0
)
),
# cps.ma.normal input start
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsmanormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# narms
sliderInput(
"narmscpsmanormal",
narmstext,
value = 3,
min = 2,
max = 10,
step = 1
),
# nclusters
textInput("nclusterscpsmanormal",
simnclusterstext,
value = "10, 10, 10"),
bsTooltip(
"nclusterscpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
# nsubjects
textInput("nsubjectscpsmanormal",
simnsubjectstext,
value = "20, 20, 20"),
bsTooltip(
"nsubjectscpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
# means
textInput(
"meanscpsmanormal",
"Expected absolute treatment effect for each arm (means)",
"22.0, 21.0, 22.5"
),
bsTooltip(
"meanscpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
# ICC
textInput("ICCcpsmanormal",
analyticICCtext,
value = NULL),
bsTooltip(
"ICCcpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
# variance params
textInput("sigma_sqcpsmanormal",
analyticsigma_sqtext,
value = "1, 1, 1"),
bsTooltip(
"sigma_sqcpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
textInput("sigma_b_sqcpsmanormal",
analyticsigma_b_sqtext,
value = "1, 1, 1"),
bsTooltip(
"sigma_b_sqcpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
#other choices
selectInput(
"multi_p_methodcpsmanormal",
"Multiple comparisons adjustment",
choices = c(
"holm",
"hochberg",
"hommel",
"bonferroni",
"BH",
"BY",
"fdr",
"none"
),
selected = "bonferroni",
multiple = FALSE
),
checkboxInput("tdistcpsmanormal", "Use t-distribution", value = FALSE)
),
# cpa.ma.binary (no method)
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Binary' && input.meth == 'Analytic'",
h4(id = "noMethod", "No method exists. Use the simulation option instead."),
tags$style(HTML("#noMethod{color: #d30000;}"))
),
# cps.ma.binary input start
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsmabinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# narms
sliderInput(
"narmscpsmabinary",
narmstext,
value = 3,
min = 2,
max = 10,
step = 1
),
# nclusters
textInput("nclusterscpsmabinary",
simnclusterstext,
value = "10, 10, 10"),
bsTooltip(
"nclusterscpsmabinary",
delimtext,
'right',
options = list(container = "body")
),
# nsubjects
textInput("nsubjectscpsmabinary",
simnsubjectstext,
value = "20, 20, 20"),
bsTooltip(
"nsubjectscpsmabinary",
delimtext,
'right',
options = list(container = "body")
),
# probs
textInput(
"probscpsmabinary",
"Treatment effect probabilities for each arm (probs)",
"0.30, 0.4, 0.5"
),
bsTooltip(
"probscpsmabinary",
delimtext,
'right',
options = list(container = "body")
),
#variance params
textInput("sigma_b_sqcpsmabinary",
analyticsigma_b_sqtext,
value = "0.1, 0.1, 0.1"),
bsTooltip(
"sigma_b_sqcpsmabinary",
delimtext,
'right',
options = list(container = "body")
),
# other options
selectInput(
"multi_p_methodcpsmabinary",
"Multiple comparisons adjustment",
choices = c(
"holm",
"hochberg",
"hommel",
"bonferroni",
"BH",
"BY",
"fdr",
"none"
),
selected = "bonferroni",
multiple = FALSE
),
checkboxInput("tdistcpsmabinary", "Use t-distribution", value = FALSE)
),
# cpa.ma.count (no method)
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Count' && input.meth == 'Analytic'",
h4(id = "noMethod", "No method exists. Use the simulation option instead.")
),
# cps.ma.count input start
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsmacount",
simnsimtext,
value = 10,
max = 500000,
min = 0
),
# narms
sliderInput(
"narmscpsmacount",
narmstext,
value = 3,
min = 2,
max = 10,
step = 1
),
# nclusters
textInput("nclusterscpsmacount",
simnclusterstext,
value = "10, 10, 10"),
bsTooltip(
"nclusterscpsmacount",
delimtext,
'right',
options = list(container = "body")
),
# nsubjects
textInput("nsubjectscpsmacount",
simnsubjectstext,
value = "20, 20, 20"),
bsTooltip(
"nsubjectscpsmacount",
delimtext,
'right',
options = list(container = "body")
),
# counts
textInput(
"countscpsmacount",
"Expected count outcome for each arm (counts)",
"30, 35, 70"
),
bsTooltip(
"countscpsmacount",
"Mean event per unit time; comma delimited",
'right',
options = list(container = "body")
),
# variance params
textInput("sigma_b_sqcpsmacount",
analyticsigma_b_sqtext,
value = "1, 1.2, 1.9"),
bsTooltip(
"sigma_b_sqcpsmacount",
delimtext,
'right',
options = list(container = "body")
),
# other options
selectInput(
"multi_p_methodcpsmacount",
"Multiple comparisons adjustment",
choices = c(
"holm",
"hochberg",
"hommel",
"bonferroni",
"BH",
"BY",
"fdr",
"none"
),
selected = "bonferroni",
multiple = FALSE
),
checkboxInput("tdistcpsmacount", "Use t-distribution", value = FALSE)
),
# cpa.did.normal input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Normal' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpadidnormal", analyticnclusterstext, value = 10),
# nsubjects
numericInput("nsubjectscpadidnormal",
analyticnsubjectstext,
value = 20),
# d
numericInput("dcpadidnormal", dtext, value = 1.02),
# ICC
numericInput(
"ICCcpadidnormal",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
),
# variance params
numericInput("rho_ccpadidnormal",
rho_ctext,
value = 0),
numericInput("rho_scpadidnormal",
rho_stext,
value = 0),
numericInput("vartcpadidnormal", varttext, value = 3)
),
# cps.did.normal input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsdidnormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpsdidnormal", refsimnclusterstext, value = 6),
# nsubjects
textInput("nsubjects1cpsdidnormal",
refsimnsubjectstext,
value = "120"),
bsTooltip(
"nsubjects1cpsdidnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu
numericInput(
"delta_mucpsdidnormal",
"Reference arm expected change from baseline to followup (delta_mu)",
value = 0
),
bsTooltip(
"delta_mucpsdidnormal",
"Usually 0",
'right',
options = list(container = "body")
),
# variance params
numericInput(
"sigma_sqcpsdidnormal",
refsigma_sqtext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq01cpsdidnormal",
didsigma_b_sqpretext,
value = 0.1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq11cpsdidnormal",
didsigma_b_sqposttext,
value = 0.1,
step = 0.001,
min = 0
),
### TREATMENT ARM
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsdidnormal", treatsimnclusterstext, value = 6),
# nsubjects
textInput("nsubjects2cpsdidnormal",
treatsimnsubjectstext,
value = "120"),
bsTooltip(
"nsubjects2cpsdidnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu
numericInput(
"delta_mu2cpsdidnormal",
"Treatment arm expected change from baseline to followup (delta_mu2)",
value = 0.48
),
# variance params
numericInput(
"sigma_sq2cpsdidnormal",
treatsigma_sqtext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq02cpsdidnormal",
didsigma_b_sqpretext,
value = 0.1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq12cpsdidnormal",
didsigma_b_sqposttext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.did.binary input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Binary' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpadidbinary", analyticnclusterstext, value = 33),
# nsubjects
numericInput("nsubjectscpadidbinary",
analyticnsubjectstext,
value = 50),
# d
numericInput("dcpadidbinary", dtext, value = 0.1),
# p
numericInput(
"pcpadidbinary",
"Mean post-test expected proportion (p)",
value = 0.5
),
# variance parameters
numericInput(
"ICCcpadidbinary",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
),
numericInput("rho_ccpadidbinary",
rho_ctext,
value = 0.3),
numericInput("rho_scpadidbinary",
rho_stext,
value = 0.7)
),
# cps.did.binary input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsdidbinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpsdidbinary", refsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects1cpsdidbinary",
refsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects1cpsdidbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p1t0
numericInput(
"p1t0cpsdidbinary",
"Reference arm baseline proportion (p1t0)",
value = 0.1,
step = 0.001,
min = 0
),
# p1t1
numericInput(
"p1t1cpsdidbinary",
"Reference arm expected outcome proportion (p1t1)",
value = 0.1,
step = 0.001,
min = 0
),
# variance param
numericInput(
"sigma_b_sq01cpsdidbinary",
didsigma_b_sqpretext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq11cpsdidbinary",
didsigma_b_sqposttext,
value = 1,
step = 0.001,
min = 0
),
## TREATMENT VALUES
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsdidbinary", treatsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects2cpsdidbinary",
treatsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsdidbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p2t0
numericInput(
"p2t0cpsdidbinary",
"Reference arm baseline proportion (p2t0)",
value = 0.2,
step = 0.001,
min = 0
),
# p2t1
numericInput(
"p2t1cpsdidbinary",
"Reference arm expected outcome proportion (p2t1)",
value = 0.5,
step = 0.001,
min = 0
),
#variance param
numericInput(
"sigma_b_sq02cpsdidbinary",
didsigma_b_sqpretext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq12cpsdidbinary",
didsigma_b_sqposttext,
value = 1,
step = 0.001,
min = 0
)
),
# cpa.did.count (no method)
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Count' && input.meth == 'Analytic'",
h4(id = "noMethod", "No method exists. Use the simulation option instead.")
),
# cps.did.count input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsdidcount",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpsdidcount", refsimnclusterstext, value = 7),
# nsubjects
textInput("nsubjects1cpsdidcount",
refsimnsubjectstext,
value = "9"),
bsTooltip(
"nsubjects1cpsdidcount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c1t0
numericInput(
"c1t0cpsdidcount",
"Reference arm baseline count (c1t0)",
value = 4,
step = 1,
min = 0
),
# c1t1
numericInput(
"c1t1cpsdidcount",
"Reference arm expected outcome count (c1t1)",
value = 5,
step = 1,
min = 0
),
#variance param
numericInput(
"sigma_b_sq01cpsdidcount",
didsigma_b_sqpretext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq11cpsdidcount",
didsigma_b_sqposttext,
value = 0.5,
step = 0.001,
min = 0
),
## Treatment VALUES
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsdidcount", treatsimnclusterstext, value = 7),
# nsubjects
textInput("nsubjects2cpsdidcount",
treatsimnsubjectstext,
value = "9"),
bsTooltip(
"nsubjects2cpsdidcount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c2t0
numericInput(
"c2t0cpsdidcount",
"Treatment arm baseline count (c1t0)",
value = 4,
step = 1,
min = 0
),
# c2t1
numericInput(
"c2t1cpsdidcount",
"Treatment arm expected outcome count (c1t1)",
value = 8,
step = 1,
min = 0
),
#variance param
numericInput(
"sigma_b_sq02cpsdidcount",
didsigma_b_sqpretext,
value = 0.5,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq12cpsdidcount",
didsigma_b_sqposttext,
value = 0.8,
step = 0.001,
min = 0
),
),
# cpa.sw.normal input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Normal' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpaswnormal", nclustersswtext, value = 8),
# nsubjects
numericInput("nsubjectscpaswnormal",
analyticnsubjectstext,
value = 8),
# ntimes
numericInput(
"ntimescpaswnormal",
ntimestext,
value = 3,
min = 0,
step = 1
),
# d
numericInput("dcpaswnormal", dtext, value = 1.5),
# variance params
numericInput(
"ICCcpaswnormal",
analyticICCtext,
value = 0.01,
step = 0.01,
min = 0,
max = 1
),
numericInput("rho_ccpaswnormal",
rho_ctext,
value = 0.8),
numericInput("rho_scpaswnormal",
rho_stext,
value = 0),
numericInput("vartcpaswnormal", varttext, value = 16)
),
# cps.sw.normal input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsswnormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# steps
numericInput("stepscpsswnormal", stepstext, value = 3),
# nclusters
numericInput("nclusterscpsswnormal", nclustersswtext, value = 12),
# nsubjects
textInput("nsubjectscpsswnormal",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjectscpsswnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# expected outcomes
numericInput("mu0cpsswnormal", mu0text, value = 1.4),
numericInput("mu1cpsswnormal", mu1text, value = 3.2),
# variance params
numericInput(
"sigma_sqcpsswnormal",
analyticsigma_sqtext,
value = 0.01,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sqcpsswnormal",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.sw.binary input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Binary' && input.meth == 'Analytic'",
# steps
numericInput("stepscpaswbinary", stepstext, value = 3),
#nclusters
numericInput("nclusterscpaswbinary", nclustersswtext, value = 9),
# nsubjects
numericInput("nsubjectscpaswbinary",
analyticnsubjectstext,
value = 20),
# estimated outcomes
numericInput(
"timeEffectcpaswbinary",
"Estimated time effect (timeEffect)",
value = 0
),
numericInput("p0cpaswbinary", p0text, value = 0.2),
numericInput("p1cpaswbinary", p1text, value = 0.31),
# ICC
numericInput(
"ICCcpaswbinary",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
)
),
# cps.sw.binary input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsswbinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# steps
numericInput("stepscpsswbinary", stepstext, value = 3),
# nclusters
numericInput("nclusterscpsswbinary", nclustersswtext, value = 12),
# nsubjects
textInput("nsubjectscpsswbinary",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjectscpsswbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
numericInput(
"p0cpsswbinary",
p0text,
value = 0.1,
step = 0.001,
min = 0,
max = 1
),
numericInput(
"p1cpsswbinary",
p1text,
value = 0.5,
step = 0.001,
min = 0,
max = 1
),
numericInput(
"sigma_b_sqcpsswbinary",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.sw.count input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Count' && input.meth == 'Analytic'",
#steps
numericInput("stepscpaswcount", stepstext, value = 3),
# nclusters
numericInput("nclusterscpaswcount", nclustersswtext, value = 12),
# nsubjects
numericInput("nsubjectscpaswcount",
analyticnsubjectstext,
value = 20),
# ICC
numericInput(
"ICCcpaswcount",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
),
numericInput(
"lambda1cpaswcount",
"Baseline rate for outcome of interest (lambda1)",
value = 0.65
),
numericInput("RRcpaswcount", "Intervention relative risk (RR)", value = 0.6)
),
# cps.sw.count input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsswcount",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# steps
numericInput("stepscpsswcount", stepstext, value = 3),
# nclusters
numericInput("nclusterscpsswcount", nclustersswtext, value = 12),
# nsubjects
textInput("nsubjectscpsswcount",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjectscpsswcount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
#est outcomes
numericInput(
"c0cpsswcount",
c0text,
value = 5,
step = 1,
min = 0
),
numericInput(
"c1cpsswcount",
c1text,
value = 7,
step = 1,
min = 0
),
# variance params
numericInput(
"sigma_b_sqcpsswcount",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.irgtt.normal input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Normal' && input.meth == 'Analytic'",
numericInput("dcpairgttnormal", dtext, value = 1.03),
## unclustered arm VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# nsubjects
numericInput(
"ncontrolscpairgttnormal",
ncontrolsirgttunclusttext,
value = 40,
step = 1,
min = 0
),
# varr
numericInput(
"varrcpairgttnormal",
"Unclustered arm residual variance (varr)",
value = 0.2,
step = 0.001,
min = 0
),
## Treatment VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
# nclusters
numericInput("nclusterscpairgttnormal",
nclustersswtext,
value = 8),
# nsubjects
numericInput("nsubjectscpairgttnormal",
analyticnsubjectstext,
value = 15),
# variance params
numericInput(
"vareicpairgttnormal",
"Subject-level random error variance (varei)",
value = 0.5,
step = 0.001,
min = 0
),
numericInput(
"varucpairgttnormal",
"Cluster-level random effect variance (varu)",
value = 1,
step = 0.001,
min = 0
)
),
# cps.irgtt.normal input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsirgttnormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## Unclustered VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# nsubjects
numericInput("nsubjectscpsirgttnormal",
nsubjectsirgttunclusttext,
value = 100),
# mu
numericInput("mucpsirgttnormal", "Unclustered arm expected mean (mu)", value = 1.1),
# sigma_sq
numericInput(
"sigma_sqcpsirgttnormal",
analyticsigma_sqtext,
value = 0.1,
step = 0.001,
min = 0
),
## Clustered VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
numericInput("nclusterscpsirgttnormal",
simnclusterstext,
value = 8),
# nsubjects
textInput("nsubjects2cpsirgttnormal",
simnsubjectstext,
value = "10"),
bsTooltip(
"nsubjects2cpsirgttnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu2
numericInput("mu2cpsirgttnormal", "Clustered arm expected mean (mu2)", value = 1.5),
# variance params
numericInput(
"ICC2cpsirgttnormal",
analyticICCtext,
value = NA,
step = 0.01,
min = 0,
max = 1
),
numericInput(
"sigma_sq2cpsirgttnormal",
analyticsigma_sqtext,
value = 0.2,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq2cpsirgttnormal",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.irgtt.binary input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Binary' && input.meth == 'Analytic'",
## unclustered VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# ncontrols
numericInput(
"ncontrolscpairgttbinary",
ncontrolsirgttunclusttext,
value = 200,
step = 1,
min = 0
),
bsTooltip(
"ncontrolscpairgttbinary",
"(in the unclustered arm)",
'right',
options = list(container = "body")
),
# p1
numericInput(
"p1cpairgttbinary",
"Unclustered arm estimated proportion (p1)",
value = 0.1,
step = 0.001,
min = 0,
max = 1
),
## Clustered VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
# nclusters
numericInput("nclusterscpairgttbinary", nclustersswtext, value = 10),
# nsubjects
numericInput("nsubjectscpairgttbinary",
analyticnsubjectstext,
value = 20),
# p2
numericInput(
"p2cpairgttbinary",
"Clustered arm estimated proportion (p2)",
value = 0.21,
step = 0.001,
min = 0,
max = 1
),
numericInput(
"ICCcpairgttbinary",
analyticICCtext,
value = 0.01,
step = 0.01,
min = 0,
max = 1
),
checkboxInput(
"decreasecpairgttbinary",
"Intervention probability < control probability",
value = FALSE
)
),
# cps.irgtt.binary input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsirgttbinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## Unclustered VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# nsubjects
numericInput("nsubjectscpsirgttbinary",
nsubjectsirgttunclusttext,
value = 200),
# p1
numericInput(
"p1cpsirgttbinary",
"Unclustered arm estimated proportion (p1)",
value = 0.1,
step = 0.001,
min = 0,
max = 1
),
## Clustered VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
# nclusters
numericInput("nclusterscpsirgttbinary",
simnclusterstext,
value = 10),
# nsubjects
textInput("nsubjects2cpsirgttbinary",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsirgttbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p2
numericInput(
"p2cpsirgttbinary",
"Clustered arm estimated proportion (p2)",
value = 0.21,
step = 0.001,
min = 0,
max = 1
),
# variance params
numericInput(
"sigma_b_sq2cpsirgttbinary",
analyticsigma_b_sqtext,
value = 0.02,
step = 0.001,
min = 0
)
),
# cpa.irgtt.count (no method)
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Count' && input.meth == 'Analytic'",
h4(id = "noMethod", "No method exists. Use the simulation option instead.")
),
# cps.irgtt.count input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsirgttcount",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## Unclustered VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# nsubjects
numericInput("nsubjectscpsirgttcount",
nsubjectsirgttunclusttext,
value = 200),
# c1
numericInput(
"c1cpsirgttcount",
"Unclustered arm estimated count (c1)",
value = 5,
step = 1,
min = 0
),
## Clustered VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
# nclusters
numericInput("nclusterscpsirgttcount",
simnclusterstext,
value = 10),
# nsubjects
textInput("nsubjects2cpsirgttcount",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsirgttcount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c2
numericInput(
"c2cpsirgttcount",
"Clustered arm estimated count (c2)",
value = 7,
step = 1,
min = 0
),
# variance param
numericInput(
"sigma_b_sq2cpsirgttcount",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
)
),
#end of values that can be reset with the restore defaults button
actionButton(
"button",
"Estimate Power",
icon = icon("arrow-circle-right"),
width = '100%',
style = "color: #fff; background-color: #337ab7; border-color: #2e6da4"
),
actionButton('cancel', 'Cancel'),
checkboxInput("more", "Show advanced options", value = FALSE),
conditionalPanel(
"input.more == true",
bookmarkButton("Save App State"),
sliderInput(
"alpha",
"Significance level (alpha)",
value = 0.05,
min = 0.01,
max = 0.1,
step = 0.02
),
checkboxInput("verbose", "Show verbose results", value = FALSE),
checkboxInput("debug", "Show debug/diagnostics (advanced)", value = FALSE)
),
conditionalPanel(
"input.more == true && input.meth == 'Simulation'",
checkboxInput("timelimitOverride", "Allow unlimited calculation time", value = FALSE),
checkboxInput("lowPowerOverride", "Allow completion when power is < 0.5", value = FALSE),
checkboxInput(
"poorFitOverride",
"Allow completion when model fit is poor",
value = FALSE
),
textInput("optmethod", "Specify an optimization method", value = "NLOPT_LN_NELDERMEAD"),
numericInput(
"seed",
"Set the seed (for repeatability)",
value = NA,
step = 1
)
),
conditionalPanel(
"input.more == true",
actionButton(
"restoreDefault",
"Restore default parameters",
width = '100%',
style = "color: #fff; background-color: #337ab7; border-color: #2e6da4"
),
actionButton("reload", "Reset all", icon = icon("trash-alt"))
)
),
########################
# Tabs start
########################
mainPanel(tabsetPanel(
tabPanel(
"Results",
conditionalPanel(
"input.dismissMsg == false && input.dismissMsg2 == false && input.more == false",
wellPanel(
HTML(
"<p>Note: If the estimated calculation time is longer than 3 minutes,
the process will not run unless you override the time limit
under advanced options.</p>"
),
checkboxInput("dismissMsg", "dismiss this message", value = FALSE)
)
),
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Binary' &&
input.meth == 'Analytic' && input.stepscpaswbinary > 3 &&
input.dismissMsgCrossover == false",
wellPanel(
HTML(
"<p>Note: Including time effects or many crossover steps may
substantially increase calculation time. </p>"
),
checkboxInput("dismissMsgCrossover", "dismiss this message", value = FALSE)
)
),
verbatimTextOutput("CRTpower", placeholder = TRUE),
conditionalPanel("input.meth == 'Simulation'",),
#### DEBUG ACCESS PANEL START #####
conditionalPanel(
"input.debug == true",
textInput("debugSearch", "Search input", value = ""),
tableOutput("show_inputs")
)
#### DEBUG ACCESS PANEL END #####
),
tabPanel(
"Graphs",
conditionalPanel(
"input.dismissMsg == false && input.dismissMsg2 == false && input.more == false",
wellPanel(
HTML(
"<p>Note: If the estimated calculation time is longer than 3 minutes,
the process will not run unless you override the time limit
under advanced options.</p>"
),
checkboxInput("dismissMsg2", "dismiss this message", value = FALSE)
)
),
selectInput("axisname",
"Y-axis name",
choices = c("nclusters", "nsubjects")),
plotOutput("graphic", click = "click"),
tableOutput("dp"),
actionButton("cleargraph", "Clear Data", icon = icon("trash-alt")),
tags$style(type = 'text/css', "button#cleargraph { margin-top: 100px; }")
),
tabPanel(
"Parameters",
h5(
id = "helpnote",
"Note: This tab shows help for the
clusterPower functions. For help with this app, see
the Help tab."
),
shinyjs::hidden(
textInput("fxnName", "clusterPower function name", value = "cpa.normal")
),
wellPanel(
HTML(
"<p>This table shows the values that the Shiny app passes
to the R functions.</p>
<p>Note: for more advanced features, see the"
),
tags$a(
" clusterPower R package.",
href = "https://cran.r-project.org/web/packages/clusterPower/clusterPower.pdf",
target = "_blank"
)
),
fluidRow(
column(width = 6, tableOutput("tracker")),
column(width = 6)
),
downloadButton(
"downloadData",
"Download this table (.csv)",
icon = icon("file-download"),
style = "color: #fff; background-color: #337ab7; border-color: #2e6da4"
),
actionButton("cleargraph2", "Clear Data", icon = icon("trash-alt")),
tags$style(type = 'text/css', "button#cleargraph2 { margin-left: 250px; }")
),
tabPanel(
"Help",
h5(
id = "helpnote",
"Note: This tab shows help for this app.
For help with the clusterPower functions, see the Parameters tab."
),
tags$style(HTML("#helpnote{color: #337ab7;}")),
HTML(
"<p>To use the calculator, select the trial type, outcome distribution, and calculation method.
Then enter values for the quantities that appear below. When complete, select the ESTIMATE POWER button.</p>"
),
HTML("<h3>Getting started</h3>"),
HTML(
"<p>The clusterPower package is intended to perform power calculations for many of the most common
randomized controlled trial (RCT) designs. This app does not allow the user to access all of the
functions available in the clusterPower R package, such as calculating the numbers of clusters or
subjects needed to obtain a specific power, returning the raw simulated datasets, or viewing the
results of each model fit. For these functions, use clusterPower with the R console rather than
from within the applet.</p>
<p>The first and most important step for using this app is to choose the appropriate experimental
design and outcome distribution for your RCT. For more on this topic, consult the "
),
tags$a(
"clusterPower vignette.",
href = "https://cran.r-project.org/web/packages/clusterPower/vignettes/clusterpower.html",
target = "_blank"
),
HTML(
"To return to this page, click the back or reload button at the top of your browser window.</p>
<h4>Choosing a distribution</h4>
<p>After you have selected the RCT type using the pulldown menu, select the outcome distribution using
the next pulldown menu. The options here include normal, binary, and count distributions. Use normal
distribution when your measurement of interest is a numeric value. This can include measurements like
descriptive variables such as weights, lab results, density, etc. Choose binary distribution if your
outcome is a yes/no type of response metric, which would be found in studies with outcomes
represented by exactly two choices (sometimes qualitative), such as survived/deceased, uninfected/infected,
or participated/withdrew. Choose count when the outcome has more than two possible outcomes, such as
uninfected/infected/recovered/died.</p>
<h4>Choosing a method</h4>
<p>The user can choose the calculation method using the 'Method' dropdown menu. The choices are analytical
or simulation. Analytical methods have the advantage that calculations are sometimes faster than
simulated methods, but many make assumptions about variance or balance in design that may sacrifice
the accuracy of the power estimation. Simulated methods take longer to run but are more flexible because
they make fewer assumptions. Furthermore, analytical methods don't exist for all RCT types, meaning that in
those cases the simulation approach may be the user's only option. However, analytical and simulated
power estimation will likely produce similar results.</p>
<h4>Parameters</h4>
<p>Depending on the user's choices as outlined in the previous sections, different parameter entry
options will appear in the left-hand panel. All of these include the number of observations (or
subjects, depending on design requirements), and the number of clusters in each arm. Other options
include variables representing the expected outcomes for each arm, and those representing measures
of variablity among and within clusters. For simulated methods, the user can also supply a number
of simulations they would like to use for calculation.</p>
<p>Here the user may want to consult the 'Parameters' tab option, which displays the verbatim
parameter values that the app passes to the clusterPower backend as they are entered by the user.
This table also shows the internal argument names for each parameter.</p>
<h4>Advanced options</h4>
<p>The checkbox near the bottom of the left-hand panel opens the Advanced Options for the app. For
all analysis types, these options include options to bookmark the app state, restore the default
parameters, or reset (reload) the app. Users can also adjust the significance cutoff value alpha
using the slider. Typically this value is set to 0.05, which is the default for this app.</p>
<p>For the simulation methods, there are 3-4 additional controls. These include:</p>
<p>1) Time limit override: Simulation methods sometimes take a long time to complete. Each
operation will produce a time estimate based on a few model fits. If the estimate is longer than
2 minutes, the function will stop and return the estimated time to produce the desired fits.
Unchecking this option allows the app to run indefinitely, so users can remove the time limit
constraint if they choose.</p>
<p>2) Low power override: The simulated methods also have a built-in option to stop fitting
models if the calculated power is lower than 0.5. Generally, 0.8 is the ideal power target
for RCT power estimation, so estimations that are very low trigger the alogrithm to stop
fitting so that the user doesn't have to wait for the run to finish. However, the low power
error can be overridden by selecting the low power override option.</p>
<p>3) Poor fit override: When more than 25% of models fail to converge, the default app state
will stop the calculation with an error, again to prevent the user from waiting a long time
for uninformative results. Lack of convergence generally is an indication that the data is
not a good fit for the models. This is ideally addressed by adjusting the model parameters.
However, if the user wants to allow the procedure to run despite lack of convergence, the
poor fit override option will override this error. Use this option with caution, as models
lacking convergence may produce unreliable estimates.</p>
<p>4) Some simulated methods allow the user to specify an optimizer, which can sometimes
address convergence issues as an alternative to overriding the poor fit checks or excluding
non-convergent models.</p>
<h3>Results</h3>
<p>After selecting the desired parameters, submit the job by clicking the Estimate Power button
near the bottom of the screen. When complete, results will appear on the Results tab. Please
keep in mind that calculations may take up to 2 minutes to complete, unless the user has chosen
to override the time limit, in which case the wait time may be longer. Wait times vary depending
on the CRT design and complexity of the resulting model, whether the method is analytic or
simulation, and the amount of RAM available in the host computer. The Results tab shows the power
calculation resulting from the most recent clusterPower run, although consecutive runs are logged
unless the cached data is manually cleared, or the CRT type, method, or distribution is
changed by the user.</p>
<h3>Graphics</h3>
<p> As mentioned in the previous section, consecutive runs are logged
unless the cached data is manually cleared, or the CRT type, method, or distribution is
changed by the user. On the Graphics tab, the user can graph any of the user-selected
parameters against the resulting power estimate using the drop-down menu. Exact coordinates
for each point can be obtained by clicking on the point of interest. If the user would like
to clear the graph manually, the Clear Data button at the bottom of the panel will clear the
Results and Parameters tabs to their original state.</p>
<h3>Parameters</h3>
<p>On the parameters tab, consecutive clusterPower runs are logged until the data is cleared
or the user selects a different CRT type, method, or distribution. Parameters are shown
according to their argument names when passed to the clusterPower function. To learn more
about each parameter, select the 'Tell Me More About This' link near the top of the panel
to open a pop-up window showing the documentation for the clusterPower function in use.</p>
<h3>Note</h3>
<p>App created by Alexandria Sakrejda, Jon Moyer, and Ken Kleinman; support from NIGMS grant R01GM121370.
Please contact ken.kleinman@gmail.com with any feedback.</p>"
)
)
)) # Tabs end
)
)
######################################
# SERVER
######################################
server <- function(input, output, session) {
disable("cancel")
out1 <- reactiveValues(power = NULL)
logargs <- reactiveValues(tab = NULL)
# Reload the app
observeEvent(input$reload, {
session$reload()
})
# Restore default values
observeEvent(input$restoreDefault, {
shinyjs::reset("allValues")
})
#change text input to numeric
textToNum <- function(x) {
result <- as.numeric(unlist(strsplit(x, split = ",")))
return(result)
}
#make some helpful fxns to extract arg names
updateArgs <- function(fxnName) {
argMatchResult <-
c(
clusterPower::argMatch(fxnName, justNames = TRUE),
"lowPowerOverride",
"poorFitOverride",
"timelimitOverride",
"power",
"seed",
"optmethod"
)
argNames <-
c("nsubjects",
"nclusters",
"alpha",
dplyr::intersect(argMatchResult,
names(formals(fxnName))))
arghelper <- function(argname) {
x <- paste0("input$", argname)
return(x)
}
holder <- list()
for (i in 1:length(argNames)) {
holder[[i]] <- arghelper(argNames[i])
}
names(holder) <- argNames
return(holder)
}
printresult <- function(fxnName) {
x <- rlang::exec(fxnName, !!!updateArgs(fxnName))
return(x)
}
#end of fxns to extract argument names
# update help documentation and params table when function is selected
observe({
if (input$type == 'Parallel' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.count")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.count")
shinyjs::show(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.ma.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.ma.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.ma.binary")
shinyjs::hide(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.ma.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.ma.count")
shinyjs::hide(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.ma.count")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.did.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.did.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.did.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.did.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.did.count")
shinyjs::hide(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.did.count")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.sw.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.sw.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.sw.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.sw.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.sw.count")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.sw.count")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.irgtt.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.irgtt.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.irgtt.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.irgtt.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.irgtt.count")
shinyjs::hide(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.irgtt.count")
shinyjs::show(id = "button")
}
}) # end update help documentation and params table when function is selected
# call the clusterPower functions
observeEvent(input$button, {
disable("button")
enable("cancel")
prog <- Progress$new(session)
prog$set(message = "Analysis in progress",
value = NULL)
isolate({
q <- reactiveValuesToList(input)
})
if (input$type == 'Parallel' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <-
cpa.normal(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpanormal,
nsubjects = q$nsubjectscpanormal,
sigma_sq = q$sigma_sqcpanormal,
sigma_b_sq = q$sigma_b_sqcpanormal,
CV = q$CVcpanormal,
d = q$dcpanormal,
ICC = q$ICCcpanormal,
vart = q$vartcpanormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.normal(
nsim = q$nsimcpsnormal,
nclusters = c(q$nclusters1cpsnormal, q$nclusters2cpsnormal),
nsubjects = textToNum(c(
q$nsubjects1cpsnormal, q$nsubjects2cpsnormal
)),
mu = q$mucpsnormal,
mu2 = q$mu2cpsnormal,
ICC = q$ICCcpsnormal,
sigma_sq = q$sigma_sqcpsnormal,
sigma_b_sq = q$sigma_b_sqcpsnormal,
ICC2 = q$ICC2cpsnormal,
sigma_sq2 = q$sigma_sq2cpsnormal,
sigma_b_sq2 = q$sigma_b_sq2cpsnormal,
alpha = q$alpha,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
timelimitOverride = q$timelimitOverride,
lowPowerOverride = q$lowPowerOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.binary(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpabinary,
nsubjects = q$nsubjectscpabinary,
CV = q$CVcpabinary,
p1 = q$p1cpabinary,
p2 = q$p2cpabinary,
ICC = q$ICCcpabinary,
pooled = q$pooledcpabinary,
p1inc = q$p1inccpabinary,
tdist = q$tdistcpabinary
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.binary(
nsim = q$nsimcpsbinary,
nsubjects = textToNum(c(
q$nsubjects1cpsbinary, q$nsubjects2cpsbinary
)),
nclusters = c(q$nclusters1cpsbinary, q$nclusters2cpsbinary),
p1 = q$p1cpsbinary,
p2 = q$p2cpsbinary,
sigma_b_sq = q$sigma_b_sqcpsbinary,
sigma_b_sq2 = q$sigma_b_sq2cpsbinary,
alpha = q$alpha,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.count(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpacount,
nsubjects = q$nsubjectscpacount,
r1 = q$r1cpacount,
r2 = q$r2cpacount,
CVB = q$CVBcpacount,
r1inc = q$r1inccpacount
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.count(
nsim = q$nsimcpscount,
nsubjects = textToNum(c(
q$nsubjects1cpscount, q$nsubjects2cpscount
)),
nclusters = c(q$nclusters1cpscount, q$nclusters2cpscount),
c1 = q$c1cpscount,
c2 = q$c2cpscount,
sigma_b_sq = q$sigma_b_sqcpscount,
sigma_b_sq2 = q$sigma_b_sq2cpscount,
alpha = q$alpha,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.ma.normal(
alpha = q$alpha,
power = q$power,
narms = q$narmscpamanormal,
nclusters = q$nclusterscpamanormal,
nsubjects = q$nsubjectscpamanormal,
vara = q$varacpamanormal,
varc = q$varccpamanormal,
vare = q$varecpamanormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.ma.normal(
nsim = q$nsimcpsmanormal,
nsubjects = textToNum(q$nsubjectscpsmanormal),
narms = q$narmscpsmanormal,
nclusters = textToNum(q$nclusterscpsmanormal),
means = textToNum(q$meanscpsmanormal),
sigma_sq = textToNum(q$sigma_sqcpsmanormal),
sigma_b_sq = textToNum(q$sigma_b_sqcpsmanormal),
alpha = q$alpha,
ICC = textToNum(q$ICCcpsmanormal),
multi_p_method = q$multi_p_methodcpsmanormal,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
tdist = q$tdistcpsmanormal,
optmethod = q$optmethod,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
answer <<- future({
val <- cpa.ma.binary()
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.ma.binary(
nsim = q$nsimcpsmabinary,
nsubjects = textToNum(q$nsubjectscpsmabinary),
narms = q$narmscpsmabinary,
nclusters = textToNum(q$nclusterscpsmabinary),
probs = textToNum(q$probscpsmabinary),
sigma_b_sq = textToNum(q$sigma_b_sqcpsmabinary),
alpha = q$alpha,
multi_p_method = q$multi_p_methodcpsmabinary,
seed = q$seed,
tdist = q$tdistcpsmabinary,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
answer <<- future({
val <- cpa.ma.count()
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.ma.count(
nsim = q$nsimcpsmacount,
nsubjects = textToNum(q$nsubjectscpsmacount),
narms = q$narmscpsmacount,
nclusters = textToNum(q$nclusterscpsmacount),
counts = textToNum(q$countscpsmacount),
sigma_b_sq = textToNum(q$sigma_b_sqcpsmacount),
alpha = q$alpha,
multi_p_method = q$multi_p_methodcpsmacount,
seed = q$seed,
tdist = q$tdistcpsmacount,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.did.normal(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpadidnormal,
nsubjects = q$nsubjectscpadidnormal,
d = q$dcpadidnormal,
ICC = q$ICCcpadidnormal,
rho_c = q$rho_ccpadidnormal,
rho_s = q$rho_scpadidnormal,
vart = q$vartcpadidnormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.did.normal(
nsim = q$nsimcpsdidnormal,
nsubjects = textToNum(
c(q$nsubjects1cpsdidnormal, q$nsubjects2cpsdidnormal)
),
nclusters = c(q$nclusters1cpsdidnormal, q$nclusters2cpsdidnormal),
delta_mu = q$delta_mucpsdidnormal,
delta_mu2 = q$delta_mu2cpsdidnormal,
sigma_sq = q$sigma_sqcpsdidnormal,
sigma_b_sq0 = c(
q$sigma_b_sq01cpsdidnormal,
q$sigma_b_sq02cpsdidnormal
),
sigma_b_sq1 = c(
q$sigma_b_sq11cpsdidnormal,
q$sigma_b_sq12cpsdidnormal
),
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.did.binary(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpadidbinary,
nsubjects = q$nsubjectscpadidbinary,
p = q$pcpadidbinary,
d = q$dcpadidbinary,
ICC = q$ICCcpadidbinary,
rho_c = q$rho_ccpadidbinary,
rho_s = q$rho_scpadidbinary
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.did.binary(
nsim = q$nsimcpsdidbinary,
nsubjects = textToNum(
c(q$nsubjects1cpsdidbinary, q$nsubjects2cpsdidbinary)
),
nclusters = c(q$nclusters1cpsdidbinary, q$nclusters2cpsdidbinary),
p1t0 = q$p1t0cpsdidbinary,
p2t0 = q$p2t0cpsdidbinary,
p1t1 = q$p1t1cpsdidbinary,
p2t1 = q$p2t1cpsdidbinary,
sigma_b_sq0 = c(
q$sigma_b_sq01cpsdidbinary,
q$sigma_b_sq02cpsdidbinary
),
sigma_b_sq1 = c(
q$sigma_b_sq11cpsdidbinary,
q$sigma_b_sq12cpsdidbinary
),
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
answer <<- future({
val <- cpa.did.count()
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.did.count(
nsim = q$nsimcpsdidcount,
nsubjects = textToNum(c(
q$nsubjects1cpsdidcount, q$nsubjects2cpsdidcount
)),
nclusters = c(q$nclusters1cpsdidcount, q$nclusters2cpsdidcount),
c1t0 = q$c1t0cpsdidcount,
c2t0 = q$c2t0cpsdidcount,
c1t1 = q$c1t1cpsdidcount,
c2t1 = q$c2t1cpsdidcount,
sigma_b_sq0 = c(
q$sigma_b_sq01cpsdidcount,
q$sigma_b_sq02cpsdidcount
),
sigma_b_sq1 = c(
q$sigma_b_sq11cpsdidcount,
q$sigma_b_sq12cpsdidcount
),
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.sw.normal(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpaswnormal,
nsubjects = q$nsubjectscpaswnormal,
ntimes = q$ntimescpaswnormal,
d = q$dcpaswnormal,
ICC = q$ICCcpaswnormal,
rho_c = q$rho_ccpaswnormal,
rho_s = q$rho_scpaswnormal,
vart = q$vartcpaswnormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.sw.normal(
alpha = q$alpha,
nsim = q$nsimcpsswnormal,
nclusters = q$nclusterscpsswnormal,
nsubjects = textToNum(q$nsubjectscpsswnormal),
steps = q$stepscpsswnormal,
mu0 = q$mu0cpsswnormal,
mu1 = q$mu1cpsswnormal,
sigma_sq = q$sigma_sqcpsswnormal,
sigma_b_sq = q$sigma_b_sqcpsswnormal,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.sw.binary(
alpha = q$alpha,
nclusters = q$nclusterscpaswbinary,
steps = q$stepscpaswbinary,
nsubjects = q$nsubjectscpaswbinary,
timeEffect = q$timeEffectcpaswbinary,
ICC = q$ICCcpaswbinary,
p1 = q$p1cpaswbinary,
p0 = q$p0cpaswbinary
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.sw.binary(
nsim = q$nsimcpsswbinary,
nsubjects = textToNum(q$nsubjectscpsswbinary),
nclusters = q$nclusterscpsswbinary,
p0 = q$p0cpsswbinary,
p1 = q$p1cpsswbinary,
steps = q$stepscpsswbinary,
sigma_b_sq = q$sigma_b_sqcpsswbinary,
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.sw.count(
lambda1 = q$lambda1cpaswcount,
RR = q$RRcpaswcount,
nclusters = q$nclusterscpaswcount,
steps = q$stepscpaswcount,
nsubjects = q$nsubjectscpaswcount,
ICC = q$ICCcpaswcount,
alpha = q$alpha
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.sw.count(
nsim = q$nsimcpsswcount,
nsubjects = textToNum(q$nsubjectscpsswcount),
nclusters = q$nclusterscpsswcount,
c0 = q$c0cpsswcount,
c1 = q$c1cpsswcount,
steps = q$stepscpsswcount,
sigma_b_sq = q$sigma_b_sqcpsswcount,
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.irgtt.normal(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpairgttnormal,
nsubjects = q$nsubjectscpairgttnormal,
ncontrols = q$ncontrolscpairgttnormal,
d = q$dcpairgttnormal,
varu = q$varucpairgttnormal,
varei = q$vareicpairgttnormal,
varr = q$varrcpairgttnormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.irgtt.normal(
nsim = q$nsimcpsirgttnormal,
nsubjects = textToNum(
c(
q$nsubjectscpsirgttnormal,
q$nsubjects2cpsirgttnormal
)
),
nclusters = q$nclusterscpsirgttnormal,
mu = q$mucpsirgttnormal,
mu2 = q$mu2cpsirgttnormal,
sigma_sq = q$sigma_sqcpsirgttnormal,
ICC2 = q$ICC2cpsirgttnormal,
sigma_sq2 = q$sigma_sq2cpsirgttnormal,
sigma_b_sq2 = q$sigma_b_sq2cpsirgttnormal,
alpha = q$alpha,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.irgtt.binary(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpairgttbinary,
nsubjects = q$nsubjectscpairgttbinary,
ncontrols = q$ncontrolscpairgttbinary,
ICC = q$ICCcpairgttbinary,
p2 = q$p2cpairgttbinary,
p1 = q$p1cpairgttbinary,
decrease = q$decreasecpairgttbinary
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.irgtt.binary(
nsim = q$nsimcpsirgttbinary,
nsubjects = textToNum(
c(
q$nsubjectscpsirgttbinary,
q$nsubjects2cpsirgttbinary
)
),
nclusters = q$nclusterscpsirgttbinary,
p1 = q$p1cpsirgttbinary,
p2 = q$p2cpsirgttbinary,
sigma_b_sq2 = q$sigma_b_sq2cpsirgttbinary,
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
answer <<- future({
val <- cpa.irgtt.count()
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.irgtt.count(
nsim = q$nsimcpsirgttcount,
nsubjects = textToNum(
c(q$nsubjectscpsirgttcount, q$nsubjects2cpsirgttcount)
),
nclusters = q$nclusterscpsirgttcount,
c1 = q$c1cpsirgttcount,
c2 = q$c2cpsirgttcount,
sigma_b_sq2 = q$sigma_b_sq2cpsirgttcount,
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
then(
answer,
onFulfilled = function(value) {
for (i in names(value)) {
out1$power[[i]] <- value[[i]]
}
out1 <<- out1
},
onRejected = function(error) {
out1$power$power <- paste0("ERROR: ", error$message)
}
)
finally(answer, function() {
prog$close()
disable("cancel")
enable("button")
})
return(NULL)
}, ignoreInit = TRUE) # end call the clusterPower functions
##################
# cancel button
observeEvent(input$cancel, {
async_pid <- answer$process$get_pid()
print(paste("Killing PID:", async_pid))
answer$process$kill()
out1$power <- NULL
disable("cancel")
enable("run1")
}, ignoreInit = TRUE)
#########################################
#### DEBUG ACCESS PANEL #####
#########################################
AllInputs <- reactive({
x <- reactiveValuesToList(input)
holderDebug <- NULL
if (sum(grepl("click", names(x))) == 1) {
x$click <- NULL
}
holderDebug <- data.frame(
names = names(isolate(x)),
values = unlist(isolate(x), use.names = FALSE),
mode = unlist(lapply(isolate(x), mode))
)
holderDebug <- holderDebug[order(holderDebug$names), ]
if (input$debugSearch != "") {
holderDebug <- holderDebug[grepl(x$debugSearch, holderDebug$names), ]
}
return(holderDebug)
})
output$show_inputs <- renderTable({
AllInputs()
})
######################################### END DEBUG table
#create the graphing table
observeEvent(req(out1$power$power), {
if (is.character(isolate(out1$power$power)) == FALSE) {
x <- reactiveValuesToList(input)
holder <- NULL
if (sum(grepl("click", names(x))) == 1) {
x$click <- NULL
}
holder <- data.frame(
argument = names(isolate(x)),
values = unlist(isolate(x), use.names = FALSE)
)
specialnames <-
dplyr::filter(holder, grepl(gsub("\\.", "", input$fxnName), argument))
specialnames$argument <-
gsub(gsub("\\.", "", input$fxnName),
"",
specialnames$argument)
specialnames <- dplyr::arrange(specialnames, argument)
if (x$meth == "Analytic") {
tab <-
rbind(specialnames,
c("alpha", isolate(input$alpha)),
c("power", round(out1$power$power, 3)))
}
if (x$meth == "Simulation") {
tab <-
rbind(
specialnames,
c("alpha", isolate(input$alpha)),
c("upper CI", round(out1$power$power$Upper.95.CI, 3)),
c("power", round(out1$power$power$Power, 3)),
c("lower CI", round(out1$power$power$Lower.95.CI, 3))
)
}
if (is.null(logargs$tab)) {
logargs$tab <- tab
} else {
tab <- dplyr::select(tab, values)
tab <- cbind.data.frame(logargs$tab, tab)
logargs$tab <- data.frame(tab, check.names = TRUE)
}
} else {
# if logargs$tab is an error, ignore it
if (!is.null(logargs$tab)) {
tab <- cbind.data.frame(logargs$tab)
logargs$tab <- data.frame(tab, check.names = TRUE)
}
}
}) # END create the graphing table
#clear the data log under certain circumstances
observeEvent(input$cleargraph, {
logargs$tab <- NULL
})
observeEvent(input$cleargraph2, {
logargs$tab <- NULL
})
observeEvent(input$fxnName, {
logargs$tab <- NULL
})
# END clear the data log under certain circumstances
# make the graph
#update the axis choices
observeEvent(input$fxnName, {
x <- reactiveValuesToList(input)
holder <- names(x)
specialnames <-
grep(gsub("\\.", "", isolate(input$fxnName)), holder, value = TRUE)
specialnames <-
gsub(gsub("\\.", "", isolate(input$fxnName)), "", specialnames)
specialnames <-
specialnames[grepl("nsim", specialnames) == FALSE]
args_ <- c("alpha", specialnames)
updateSelectInput(session, "axisname",
choices = args_)
})
plot_this <-
eventReactive(list(input$axisname, req(logargs$tab)), {
data <- data.frame(isolate(logargs$tab), check.names = TRUE)
data$values <- as.numeric(data$values)
data$argument <- as.factor(data$argument)
var <- isolate(input$axisname)
data <-
dplyr::filter(data, argument == !!var |
argument == "power")
# first remember the names
n <- data$argument
# transpose all but the first column (name)
data <- data.frame(t(data[, -1]))
colnames(data) <- n
data[, 1:2] <- apply(data[, 1:2], 2,
function(x)
as.numeric(as.character(x)))
return(data)
})
dpfun <- function(x) {
var <- isolate(input$axisname)
data <- plot_this()
fun <- function(x) {
x <- enquo(x)
sol <-
ggplot(data, aes(x = !!x, y = power)) +
geom_point(aes(colour = "fff"), size = 2.5) +
theme_minimal() + theme(legend.position = "none")
return(sol)
}
power_plot <- suppressWarnings(fun(get(var)))
if (nrow(data) > 1) {
power_plot <-
power_plot + geom_line(aes(colour = "fff"), size = 1.25) + xlab(var)
} else {
power_plot <- power_plot + xlab(var)
}
return(power_plot)
}
output$graphic <- renderPlot({
dpfun()
}, res = 96)
output$dp <- renderTable({
q <- plot_this()
nearPoints(q, isolate(input$click), xvar = isolate(input$axisname))
})
# create reactive input data table
output$tracker <-
renderTable(logargs$tab)
# Downloadable csv of reactive input data table
output$downloadData <- downloadHandler(
filename = function() {
paste("clusterPower_", isolate(input$fxnName), ".csv", sep = "")
},
content = function(file) {
write.csv(logargs$tab, file, row.names = FALSE)
}
)
# main power output
resultdisplay <- reactive({
q <- reactiveValuesToList(out1)
if (input$verbose == FALSE)
return(q$power$power)
else
return(q$power)
})
# present the output verbose/not verbose
output$CRTpower <- renderPrint(resultdisplay())
} #end of server fxn
# Run the application
shinyApp(ui = ui, server = server)
nickreich/clusterPower documentation built on Feb. 3, 2021, 6:54 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
#
# This is a Shiny web application. You can run the application by clicking
# the 'Run App' button above.
#
# Find out more about building applications with Shiny here:
#
# http://shiny.rstudio.com/
#
library(clusterPower)
library(shinythemes)
library(shiny)
library(shinyBS)
library(promises)
library(future)
library(future.callr)
library(shinyjs)
library(V8)
library(ggplot2)
plan(callr)
# labels for arguments
simnsubjectstext <- "Number of observations per cluster (nsubjects)"
refsimnsubjectstext <-
"Number of reference arm observations (nsubjects)"
treatsimnsubjectstext <-
"Number of treatment arm observations (nsubjects)"
simnclusterstext <- "Number of clusters per arm (nclusters)"
refsimnclusterstext <-
"Number of reference arm clusters (nclusters)"
treatsimnclusterstext <-
"Number of treatment arm clusters (nclusters)"
analyticnsubjectstext <- "Mean observations per cluster (nsubjects)"
analyticnclusterstext <- "Mean clusters per arm (nclusters)"
analyticICCtext <- "Intracluster correlation coefficient (ICC)"
analyticsigma_sqtext <- "Within-cluster variance (sigma_sq)"
analyticsigma_b_sqtext <- "Between-cluster variance (sigma_b_sq)"
treatsigma_sqtext <-
"Treatment arm within-cluster variance (sigma_sq)"
refsigma_b_sqtext <-
"Reference arm between-cluster variance (sigma_b_sq)"
treatsigma_b_sqtext <-
"Treatment arm between-cluster variance (sigma_b_sq)"
simnsimtext <- "Number of simulations (nsim)"
refmutext <- "Reference arm expected mean (mu)"
treatmutext <- "Treatment arm expected mean (mu2)"
refICCtext <- "Reference arm ICC (ICC)"
treatICCtext <- "Treatment arm ICC (ICC)"
refp1text <- "Reference arm estimated proportion (p1)"
treatp2text <- "Treatment arm estimated proportion (p2)"
refc1text <- "Reference arm expected count (c1)"
treatc2text <- "Treatment arm expected count (c2)"
narmstext <- "Number of arms (narms)"
delimtext <- "Note: comma delimited"
dtext <- "Expected difference in arm means (d)"
didsigma_b_sqposttext <-
"Post-treatment between-cluster variance (sigma_b_sq1)"
didsigma_b_sqpretext <-
"Pre-treatment between-cluster variance (sigma_b_sq0)"
ntimestext <- "Number of measurement time points (ntimes)"
varttext <- "Total variation of the outcome (vart)"
rho_ctext <-
"Baseline and post-test cluster-level correlation (rho_c)"
rho_stext <-
"Baseline and post-test subject-level correlation (rho_s)"
stepstext <- "Number of crossover steps (steps)"
mu0text <- "Estimated baseline mean (mu0)"
mu1text <- "Estimated post-treatment effect (mu1)"
p0text <- "Estimated baseline proportion (p0)"
p1text <- "Estimated post-treatment effect (p1)"
c0text <- "Estimated baseline proportion (c0)"
c1text <- "Estimated post-treatment effect (c1)"
refsigma_sqtext <-
"Reference arm within-cluster variance (sigma_sq)"
nsubjectsirgttunclusttext <-
"Observations in unclustered arm (nsubjects)"
ncontrolsirgttunclusttext <-
"Observations in unclustered arm (ncontrols)"
nclustersswtext <- "Number of clusters (nclusters)"
nsubjectsdelim <-
"Comma delimited. Length of nsubjects can be 1 (all equal) or equal to the number of clusters."
ui <- fluidPage(
theme = shinytheme("united"),
h1(
id = "big-heading",
"Power Estimation for Randomized Controlled Trials: clusterPower"
),
tags$style(HTML("#big-heading{color: #337ab7;}")),
shinyjs::useShinyjs(),
sidebarLayout(
sidebarPanel(
tags$head(
tags$style(type = 'text/css',
'form.well { max-height: 700px; overflow-y: auto; }')
),
selectInput(
"type",
"CRT Type",
choices = c(
"Parallel",
"Multi-Arm",
"Difference-in-Difference",
"Stepped Wedge",
"Individually-Randomized Group"
)
),
selectInput(
"dist",
"Outcome Distribution",
choices = c("Normal", "Binary", "Count")
),
selectInput("meth", "Method",
choices = c("Analytic", "Simulation")),
# Below values can be reset to defaults using the restore defaults button
div(
id = "allValues",
shinyjs::hidden(numericInput("power", "power", value = NA)),
#input for cpa.normal -------------------------------------------------------------
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Normal' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpanormal", analyticnclusterstext, value = 10),
#nsubjects
numericInput("nsubjectscpanormal",
analyticnsubjectstext,
value = 20),
# CV
numericInput("CVcpanormal", "Coefficient of variation (CV)", value = 0),
bsTooltip(
"CVcpanormal",
"When CV equals 0, all clusters are the same size.",
'right',
options = list(container = "body")
),
# d
numericInput("dcpanormal", dtext, value = 0.43),
# ICC
numericInput(
"ICCcpanormal",
analyticICCtext,
value = NA,
step = 0.01,
min = 0,
max = 1
),
# variance params
numericInput(
"sigma_sqcpanormal",
analyticsigma_sqtext,
value = 0.01,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sqcpanormal",
analyticsigma_b_sqtext,
step = 0.001,
value = 0.1,
min = 0
),
numericInput("vartcpanormal", varttext, value = NA)
),
# input for cps.normal
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsnormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
tags$style(HTML("#armHead{color: #337ab7;}")),
# nclusters
numericInput("nclusters1cpsnormal", refsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects1cpsnormal",
refsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects1cpsnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu
numericInput("mucpsnormal", refmutext, value = 2.4),
# variance params
numericInput(
"ICCcpsnormal",
refICCtext,
value = NA,
step = 0.01,
min = 0,
max = 1
),
bsTooltip(
"ICCcpsnormal",
"Intracluster correlation coefficient",
'right',
options = list(container = "body")
),
numericInput(
"sigma_sqcpsnormal",
refsigma_sqtext,
value = 0.2,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sqcpsnormal",
refsigma_b_sqtext,
value = 0.5,
step = 0.001,
min = 0
),
### TREATMENT ARM
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsnormal", treatsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects2cpsnormal",
treatsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu
numericInput("mu2cpsnormal", treatmutext, value = 1.5),
# variance params
numericInput(
"ICC2cpsnormal",
treatICCtext,
value = NA,
step = 0.01,
min = 0,
max = 1
),
bsTooltip(
"ICC2cpsnormal",
"Intracluster correlation coefficient",
'right',
options = list(container = "body")
),
numericInput(
"sigma_sq2cpsnormal",
treatsigma_sqtext,
value = 0.2,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq2cpsnormal",
treatsigma_b_sqtext,
value = 0.5,
step = 0.001,
min = 0
)
),
# cpa.binary inputs start
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Binary' && input.meth == 'Analytic'",
# nuclusters
numericInput("nclusterscpabinary", analyticnclusterstext, value = 10),
# nsubjects
numericInput("nsubjectscpabinary",
analyticnsubjectstext,
value = 20),
# CV
numericInput("CVcpabinary", "Coefficient of variation (CV)", value = 0),
# ICC
numericInput(
"ICCcpabinary",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
),
# p1 & p2
numericInput(
"p1cpabinary",
refp1text,
value = 0.1,
step = 0.001,
min = 0,
max = 1
),
numericInput(
"p2cpabinary",
treatp2text,
value = 0.24,
step = 0.001,
min = 0,
max = 1
),
# other selections
checkboxInput("pooledcpabinary", "Pooled standard error", value = FALSE),
checkboxInput("p1inccpabinary", "p1 > p2", value = FALSE),
checkboxInput("tdistcpabinary", "Use t-distribution", value = FALSE)
),
# cps.binary inputs start
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsbinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpsbinary", refsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects1cpsbinary",
refsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects1cpsbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p1
numericInput(
"p1cpsbinary",
refp1text,
value = 0.8,
step = 0.001,
min = 0,
max = 1
),
# variance param
numericInput(
"sigma_b_sqcpsbinary",
refsigma_b_sqtext,
value = 1,
step = 0.001,
min = 0
),
## TREATMENT VALUES
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsbinary", treatsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects2cpsbinary",
treatsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p2
numericInput(
"p2cpsbinary",
treatp2text,
value = 0.5,
step = 0.001,
min = 0,
max = 1
),
#variance param
numericInput(
"sigma_b_sq2cpsbinary",
treatsigma_b_sqtext,
value = 1,
step = 0.001,
min = 0
)
),
# cpa.count input starts
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Count' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpacount", analyticnclusterstext, value = 10),
# nsubjects
numericInput("nsubjectscpacount",
analyticnsubjectstext,
value = 20),
# variance param
numericInput(
"CVBcpacount",
"Between-cluster coefficient of variation (CVB)",
value = 0.01
),
# expected outcome
numericInput("r1cpacount", "Reference arm mean event rate", value = 0.2),
numericInput("r2cpacount", "Treatment arm mean event rate", value = 0.35),
# other params
checkboxInput(
"r1inccpacount",
"Intervention probability < control probability",
value = FALSE
)
),
# cps.count input starts
conditionalPanel(
"input.type == 'Parallel' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpscount",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpscount", refsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects1cpscount",
refsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects1cpscount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c1
numericInput(
"c1cpscount",
refc1text,
value = 200,
step = 1,
min = 0
),
#variance param
numericInput(
"sigma_b_sqcpscount",
refsigma_b_sqtext,
value = 0.5,
step = 0.001,
min = 0
),
## Treatment VALUES
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpscount", treatsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects2cpscount",
treatsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpscount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c2
numericInput(
"c2cpscount",
treatc2text,
value = 80,
step = 1,
min = 0
),
#variance param
numericInput(
"sigma_b_sq2cpscount",
treatsigma_b_sqtext,
value = 0.5,
step = 0.001,
min = 0
)
),
# cpa.ma.normal input starts
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Normal' && input.meth == 'Analytic'",
#narms
numericInput(
"narmscpamanormal",
narmstext,
value = 3,
max = 50,
min = 2,
step = 1
),
# nclusters
numericInput("nclusterscpamanormal", analyticnclusterstext, value = 10),
# nsubjects
numericInput("nsubjectscpamanormal",
analyticnsubjectstext,
value = 20),
# variance params
numericInput(
"varacpamanormal",
"Between-arm variance (vara)",
value = 0.02,
step = 0.001,
min = 0
),
numericInput(
"varccpamanormal",
"Between-cluster variance (varc)",
value = 0.01,
step = 0.001,
min = 0
),
numericInput(
"varecpamanormal",
"Within-cluster variance (vare)",
value = 0.1,
step = 0.001,
min = 0
)
),
# cps.ma.normal input start
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsmanormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# narms
sliderInput(
"narmscpsmanormal",
narmstext,
value = 3,
min = 2,
max = 10,
step = 1
),
# nclusters
textInput("nclusterscpsmanormal",
simnclusterstext,
value = "10, 10, 10"),
bsTooltip(
"nclusterscpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
# nsubjects
textInput("nsubjectscpsmanormal",
simnsubjectstext,
value = "20, 20, 20"),
bsTooltip(
"nsubjectscpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
# means
textInput(
"meanscpsmanormal",
"Expected absolute treatment effect for each arm (means)",
"22.0, 21.0, 22.5"
),
bsTooltip(
"meanscpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
# ICC
textInput("ICCcpsmanormal",
analyticICCtext,
value = NULL),
bsTooltip(
"ICCcpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
# variance params
textInput("sigma_sqcpsmanormal",
analyticsigma_sqtext,
value = "1, 1, 1"),
bsTooltip(
"sigma_sqcpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
textInput("sigma_b_sqcpsmanormal",
analyticsigma_b_sqtext,
value = "1, 1, 1"),
bsTooltip(
"sigma_b_sqcpsmanormal",
delimtext,
'right',
options = list(container = "body")
),
#other choices
selectInput(
"multi_p_methodcpsmanormal",
"Multiple comparisons adjustment",
choices = c(
"holm",
"hochberg",
"hommel",
"bonferroni",
"BH",
"BY",
"fdr",
"none"
),
selected = "bonferroni",
multiple = FALSE
),
checkboxInput("tdistcpsmanormal", "Use t-distribution", value = FALSE)
),
# cpa.ma.binary (no method)
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Binary' && input.meth == 'Analytic'",
h4(id = "noMethod", "No method exists. Use the simulation option instead."),
tags$style(HTML("#noMethod{color: #d30000;}"))
),
# cps.ma.binary input start
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsmabinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# narms
sliderInput(
"narmscpsmabinary",
narmstext,
value = 3,
min = 2,
max = 10,
step = 1
),
# nclusters
textInput("nclusterscpsmabinary",
simnclusterstext,
value = "10, 10, 10"),
bsTooltip(
"nclusterscpsmabinary",
delimtext,
'right',
options = list(container = "body")
),
# nsubjects
textInput("nsubjectscpsmabinary",
simnsubjectstext,
value = "20, 20, 20"),
bsTooltip(
"nsubjectscpsmabinary",
delimtext,
'right',
options = list(container = "body")
),
# probs
textInput(
"probscpsmabinary",
"Treatment effect probabilities for each arm (probs)",
"0.30, 0.4, 0.5"
),
bsTooltip(
"probscpsmabinary",
delimtext,
'right',
options = list(container = "body")
),
#variance params
textInput("sigma_b_sqcpsmabinary",
analyticsigma_b_sqtext,
value = "0.1, 0.1, 0.1"),
bsTooltip(
"sigma_b_sqcpsmabinary",
delimtext,
'right',
options = list(container = "body")
),
# other options
selectInput(
"multi_p_methodcpsmabinary",
"Multiple comparisons adjustment",
choices = c(
"holm",
"hochberg",
"hommel",
"bonferroni",
"BH",
"BY",
"fdr",
"none"
),
selected = "bonferroni",
multiple = FALSE
),
checkboxInput("tdistcpsmabinary", "Use t-distribution", value = FALSE)
),
# cpa.ma.count (no method)
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Count' && input.meth == 'Analytic'",
h4(id = "noMethod", "No method exists. Use the simulation option instead.")
),
# cps.ma.count input start
conditionalPanel(
"input.type == 'Multi-Arm' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsmacount",
simnsimtext,
value = 10,
max = 500000,
min = 0
),
# narms
sliderInput(
"narmscpsmacount",
narmstext,
value = 3,
min = 2,
max = 10,
step = 1
),
# nclusters
textInput("nclusterscpsmacount",
simnclusterstext,
value = "10, 10, 10"),
bsTooltip(
"nclusterscpsmacount",
delimtext,
'right',
options = list(container = "body")
),
# nsubjects
textInput("nsubjectscpsmacount",
simnsubjectstext,
value = "20, 20, 20"),
bsTooltip(
"nsubjectscpsmacount",
delimtext,
'right',
options = list(container = "body")
),
# counts
textInput(
"countscpsmacount",
"Expected count outcome for each arm (counts)",
"30, 35, 70"
),
bsTooltip(
"countscpsmacount",
"Mean event per unit time; comma delimited",
'right',
options = list(container = "body")
),
# variance params
textInput("sigma_b_sqcpsmacount",
analyticsigma_b_sqtext,
value = "1, 1.2, 1.9"),
bsTooltip(
"sigma_b_sqcpsmacount",
delimtext,
'right',
options = list(container = "body")
),
# other options
selectInput(
"multi_p_methodcpsmacount",
"Multiple comparisons adjustment",
choices = c(
"holm",
"hochberg",
"hommel",
"bonferroni",
"BH",
"BY",
"fdr",
"none"
),
selected = "bonferroni",
multiple = FALSE
),
checkboxInput("tdistcpsmacount", "Use t-distribution", value = FALSE)
),
# cpa.did.normal input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Normal' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpadidnormal", analyticnclusterstext, value = 10),
# nsubjects
numericInput("nsubjectscpadidnormal",
analyticnsubjectstext,
value = 20),
# d
numericInput("dcpadidnormal", dtext, value = 1.02),
# ICC
numericInput(
"ICCcpadidnormal",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
),
# variance params
numericInput("rho_ccpadidnormal",
rho_ctext,
value = 0),
numericInput("rho_scpadidnormal",
rho_stext,
value = 0),
numericInput("vartcpadidnormal", varttext, value = 3)
),
# cps.did.normal input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsdidnormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpsdidnormal", refsimnclusterstext, value = 6),
# nsubjects
textInput("nsubjects1cpsdidnormal",
refsimnsubjectstext,
value = "120"),
bsTooltip(
"nsubjects1cpsdidnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu
numericInput(
"delta_mucpsdidnormal",
"Reference arm expected change from baseline to followup (delta_mu)",
value = 0
),
bsTooltip(
"delta_mucpsdidnormal",
"Usually 0",
'right',
options = list(container = "body")
),
# variance params
numericInput(
"sigma_sqcpsdidnormal",
refsigma_sqtext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq01cpsdidnormal",
didsigma_b_sqpretext,
value = 0.1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq11cpsdidnormal",
didsigma_b_sqposttext,
value = 0.1,
step = 0.001,
min = 0
),
### TREATMENT ARM
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsdidnormal", treatsimnclusterstext, value = 6),
# nsubjects
textInput("nsubjects2cpsdidnormal",
treatsimnsubjectstext,
value = "120"),
bsTooltip(
"nsubjects2cpsdidnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu
numericInput(
"delta_mu2cpsdidnormal",
"Treatment arm expected change from baseline to followup (delta_mu2)",
value = 0.48
),
# variance params
numericInput(
"sigma_sq2cpsdidnormal",
treatsigma_sqtext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq02cpsdidnormal",
didsigma_b_sqpretext,
value = 0.1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq12cpsdidnormal",
didsigma_b_sqposttext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.did.binary input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Binary' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpadidbinary", analyticnclusterstext, value = 33),
# nsubjects
numericInput("nsubjectscpadidbinary",
analyticnsubjectstext,
value = 50),
# d
numericInput("dcpadidbinary", dtext, value = 0.1),
# p
numericInput(
"pcpadidbinary",
"Mean post-test expected proportion (p)",
value = 0.5
),
# variance parameters
numericInput(
"ICCcpadidbinary",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
),
numericInput("rho_ccpadidbinary",
rho_ctext,
value = 0.3),
numericInput("rho_scpadidbinary",
rho_stext,
value = 0.7)
),
# cps.did.binary input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsdidbinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpsdidbinary", refsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects1cpsdidbinary",
refsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects1cpsdidbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p1t0
numericInput(
"p1t0cpsdidbinary",
"Reference arm baseline proportion (p1t0)",
value = 0.1,
step = 0.001,
min = 0
),
# p1t1
numericInput(
"p1t1cpsdidbinary",
"Reference arm expected outcome proportion (p1t1)",
value = 0.1,
step = 0.001,
min = 0
),
# variance param
numericInput(
"sigma_b_sq01cpsdidbinary",
didsigma_b_sqpretext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq11cpsdidbinary",
didsigma_b_sqposttext,
value = 1,
step = 0.001,
min = 0
),
## TREATMENT VALUES
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsdidbinary", treatsimnclusterstext, value = 10),
# nsubjects
textInput("nsubjects2cpsdidbinary",
treatsimnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsdidbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p2t0
numericInput(
"p2t0cpsdidbinary",
"Reference arm baseline proportion (p2t0)",
value = 0.2,
step = 0.001,
min = 0
),
# p2t1
numericInput(
"p2t1cpsdidbinary",
"Reference arm expected outcome proportion (p2t1)",
value = 0.5,
step = 0.001,
min = 0
),
#variance param
numericInput(
"sigma_b_sq02cpsdidbinary",
didsigma_b_sqpretext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq12cpsdidbinary",
didsigma_b_sqposttext,
value = 1,
step = 0.001,
min = 0
)
),
# cpa.did.count (no method)
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Count' && input.meth == 'Analytic'",
h4(id = "noMethod", "No method exists. Use the simulation option instead.")
),
# cps.did.count input start
conditionalPanel(
"input.type == 'Difference-in-Difference' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsdidcount",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## REFERENCE VALUES
h3(id = "armHead",
"Reference Arm Parameters"),
# nclusters
numericInput("nclusters1cpsdidcount", refsimnclusterstext, value = 7),
# nsubjects
textInput("nsubjects1cpsdidcount",
refsimnsubjectstext,
value = "9"),
bsTooltip(
"nsubjects1cpsdidcount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c1t0
numericInput(
"c1t0cpsdidcount",
"Reference arm baseline count (c1t0)",
value = 4,
step = 1,
min = 0
),
# c1t1
numericInput(
"c1t1cpsdidcount",
"Reference arm expected outcome count (c1t1)",
value = 5,
step = 1,
min = 0
),
#variance param
numericInput(
"sigma_b_sq01cpsdidcount",
didsigma_b_sqpretext,
value = 1,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq11cpsdidcount",
didsigma_b_sqposttext,
value = 0.5,
step = 0.001,
min = 0
),
## Treatment VALUES
h3(id = "armHead",
"Treatment Arm Parameters"),
# nclusters
numericInput("nclusters2cpsdidcount", treatsimnclusterstext, value = 7),
# nsubjects
textInput("nsubjects2cpsdidcount",
treatsimnsubjectstext,
value = "9"),
bsTooltip(
"nsubjects2cpsdidcount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c2t0
numericInput(
"c2t0cpsdidcount",
"Treatment arm baseline count (c1t0)",
value = 4,
step = 1,
min = 0
),
# c2t1
numericInput(
"c2t1cpsdidcount",
"Treatment arm expected outcome count (c1t1)",
value = 8,
step = 1,
min = 0
),
#variance param
numericInput(
"sigma_b_sq02cpsdidcount",
didsigma_b_sqpretext,
value = 0.5,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq12cpsdidcount",
didsigma_b_sqposttext,
value = 0.8,
step = 0.001,
min = 0
),
),
# cpa.sw.normal input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Normal' && input.meth == 'Analytic'",
# nclusters
numericInput("nclusterscpaswnormal", nclustersswtext, value = 8),
# nsubjects
numericInput("nsubjectscpaswnormal",
analyticnsubjectstext,
value = 8),
# ntimes
numericInput(
"ntimescpaswnormal",
ntimestext,
value = 3,
min = 0,
step = 1
),
# d
numericInput("dcpaswnormal", dtext, value = 1.5),
# variance params
numericInput(
"ICCcpaswnormal",
analyticICCtext,
value = 0.01,
step = 0.01,
min = 0,
max = 1
),
numericInput("rho_ccpaswnormal",
rho_ctext,
value = 0.8),
numericInput("rho_scpaswnormal",
rho_stext,
value = 0),
numericInput("vartcpaswnormal", varttext, value = 16)
),
# cps.sw.normal input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsswnormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# steps
numericInput("stepscpsswnormal", stepstext, value = 3),
# nclusters
numericInput("nclusterscpsswnormal", nclustersswtext, value = 12),
# nsubjects
textInput("nsubjectscpsswnormal",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjectscpsswnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# expected outcomes
numericInput("mu0cpsswnormal", mu0text, value = 1.4),
numericInput("mu1cpsswnormal", mu1text, value = 3.2),
# variance params
numericInput(
"sigma_sqcpsswnormal",
analyticsigma_sqtext,
value = 0.01,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sqcpsswnormal",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.sw.binary input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Binary' && input.meth == 'Analytic'",
# steps
numericInput("stepscpaswbinary", stepstext, value = 3),
#nclusters
numericInput("nclusterscpaswbinary", nclustersswtext, value = 9),
# nsubjects
numericInput("nsubjectscpaswbinary",
analyticnsubjectstext,
value = 20),
# estimated outcomes
numericInput(
"timeEffectcpaswbinary",
"Estimated time effect (timeEffect)",
value = 0
),
numericInput("p0cpaswbinary", p0text, value = 0.2),
numericInput("p1cpaswbinary", p1text, value = 0.31),
# ICC
numericInput(
"ICCcpaswbinary",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
)
),
# cps.sw.binary input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsswbinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# steps
numericInput("stepscpsswbinary", stepstext, value = 3),
# nclusters
numericInput("nclusterscpsswbinary", nclustersswtext, value = 12),
# nsubjects
textInput("nsubjectscpsswbinary",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjectscpsswbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
numericInput(
"p0cpsswbinary",
p0text,
value = 0.1,
step = 0.001,
min = 0,
max = 1
),
numericInput(
"p1cpsswbinary",
p1text,
value = 0.5,
step = 0.001,
min = 0,
max = 1
),
numericInput(
"sigma_b_sqcpsswbinary",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.sw.count input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Count' && input.meth == 'Analytic'",
#steps
numericInput("stepscpaswcount", stepstext, value = 3),
# nclusters
numericInput("nclusterscpaswcount", nclustersswtext, value = 12),
# nsubjects
numericInput("nsubjectscpaswcount",
analyticnsubjectstext,
value = 20),
# ICC
numericInput(
"ICCcpaswcount",
analyticICCtext,
value = 0.05,
step = 0.01,
min = 0,
max = 1
),
numericInput(
"lambda1cpaswcount",
"Baseline rate for outcome of interest (lambda1)",
value = 0.65
),
numericInput("RRcpaswcount", "Intervention relative risk (RR)", value = 0.6)
),
# cps.sw.count input start
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsswcount",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
# steps
numericInput("stepscpsswcount", stepstext, value = 3),
# nclusters
numericInput("nclusterscpsswcount", nclustersswtext, value = 12),
# nsubjects
textInput("nsubjectscpsswcount",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjectscpsswcount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
#est outcomes
numericInput(
"c0cpsswcount",
c0text,
value = 5,
step = 1,
min = 0
),
numericInput(
"c1cpsswcount",
c1text,
value = 7,
step = 1,
min = 0
),
# variance params
numericInput(
"sigma_b_sqcpsswcount",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.irgtt.normal input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Normal' && input.meth == 'Analytic'",
numericInput("dcpairgttnormal", dtext, value = 1.03),
## unclustered arm VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# nsubjects
numericInput(
"ncontrolscpairgttnormal",
ncontrolsirgttunclusttext,
value = 40,
step = 1,
min = 0
),
# varr
numericInput(
"varrcpairgttnormal",
"Unclustered arm residual variance (varr)",
value = 0.2,
step = 0.001,
min = 0
),
## Treatment VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
# nclusters
numericInput("nclusterscpairgttnormal",
nclustersswtext,
value = 8),
# nsubjects
numericInput("nsubjectscpairgttnormal",
analyticnsubjectstext,
value = 15),
# variance params
numericInput(
"vareicpairgttnormal",
"Subject-level random error variance (varei)",
value = 0.5,
step = 0.001,
min = 0
),
numericInput(
"varucpairgttnormal",
"Cluster-level random effect variance (varu)",
value = 1,
step = 0.001,
min = 0
)
),
# cps.irgtt.normal input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Normal' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsirgttnormal",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## Unclustered VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# nsubjects
numericInput("nsubjectscpsirgttnormal",
nsubjectsirgttunclusttext,
value = 100),
# mu
numericInput("mucpsirgttnormal", "Unclustered arm expected mean (mu)", value = 1.1),
# sigma_sq
numericInput(
"sigma_sqcpsirgttnormal",
analyticsigma_sqtext,
value = 0.1,
step = 0.001,
min = 0
),
## Clustered VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
numericInput("nclusterscpsirgttnormal",
simnclusterstext,
value = 8),
# nsubjects
textInput("nsubjects2cpsirgttnormal",
simnsubjectstext,
value = "10"),
bsTooltip(
"nsubjects2cpsirgttnormal",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# mu2
numericInput("mu2cpsirgttnormal", "Clustered arm expected mean (mu2)", value = 1.5),
# variance params
numericInput(
"ICC2cpsirgttnormal",
analyticICCtext,
value = NA,
step = 0.01,
min = 0,
max = 1
),
numericInput(
"sigma_sq2cpsirgttnormal",
analyticsigma_sqtext,
value = 0.2,
step = 0.001,
min = 0
),
numericInput(
"sigma_b_sq2cpsirgttnormal",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
),
# cpa.irgtt.binary input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Binary' && input.meth == 'Analytic'",
## unclustered VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# ncontrols
numericInput(
"ncontrolscpairgttbinary",
ncontrolsirgttunclusttext,
value = 200,
step = 1,
min = 0
),
bsTooltip(
"ncontrolscpairgttbinary",
"(in the unclustered arm)",
'right',
options = list(container = "body")
),
# p1
numericInput(
"p1cpairgttbinary",
"Unclustered arm estimated proportion (p1)",
value = 0.1,
step = 0.001,
min = 0,
max = 1
),
## Clustered VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
# nclusters
numericInput("nclusterscpairgttbinary", nclustersswtext, value = 10),
# nsubjects
numericInput("nsubjectscpairgttbinary",
analyticnsubjectstext,
value = 20),
# p2
numericInput(
"p2cpairgttbinary",
"Clustered arm estimated proportion (p2)",
value = 0.21,
step = 0.001,
min = 0,
max = 1
),
numericInput(
"ICCcpairgttbinary",
analyticICCtext,
value = 0.01,
step = 0.01,
min = 0,
max = 1
),
checkboxInput(
"decreasecpairgttbinary",
"Intervention probability < control probability",
value = FALSE
)
),
# cps.irgtt.binary input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Binary' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsirgttbinary",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## Unclustered VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# nsubjects
numericInput("nsubjectscpsirgttbinary",
nsubjectsirgttunclusttext,
value = 200),
# p1
numericInput(
"p1cpsirgttbinary",
"Unclustered arm estimated proportion (p1)",
value = 0.1,
step = 0.001,
min = 0,
max = 1
),
## Clustered VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
# nclusters
numericInput("nclusterscpsirgttbinary",
simnclusterstext,
value = 10),
# nsubjects
textInput("nsubjects2cpsirgttbinary",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsirgttbinary",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# p2
numericInput(
"p2cpsirgttbinary",
"Clustered arm estimated proportion (p2)",
value = 0.21,
step = 0.001,
min = 0,
max = 1
),
# variance params
numericInput(
"sigma_b_sq2cpsirgttbinary",
analyticsigma_b_sqtext,
value = 0.02,
step = 0.001,
min = 0
)
),
# cpa.irgtt.count (no method)
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Count' && input.meth == 'Analytic'",
h4(id = "noMethod", "No method exists. Use the simulation option instead.")
),
# cps.irgtt.count input start
conditionalPanel(
"input.type == 'Individually-Randomized Group' && input.dist == 'Count' && input.meth == 'Simulation'",
# nsim
numericInput(
"nsimcpsirgttcount",
simnsimtext,
value = 100,
max = 500000,
min = 0
),
## Unclustered VALUES
h3(id = "armHead",
"Unclustered Arm Parameters"),
# nsubjects
numericInput("nsubjectscpsirgttcount",
nsubjectsirgttunclusttext,
value = 200),
# c1
numericInput(
"c1cpsirgttcount",
"Unclustered arm estimated count (c1)",
value = 5,
step = 1,
min = 0
),
## Clustered VALUES
h3(id = "armHead",
"Clustered Arm Parameters"),
# nclusters
numericInput("nclusterscpsirgttcount",
simnclusterstext,
value = 10),
# nsubjects
textInput("nsubjects2cpsirgttcount",
simnsubjectstext,
value = "20"),
bsTooltip(
"nsubjects2cpsirgttcount",
nsubjectsdelim,
'right',
options = list(container = "body")
),
# c2
numericInput(
"c2cpsirgttcount",
"Clustered arm estimated count (c2)",
value = 7,
step = 1,
min = 0
),
# variance param
numericInput(
"sigma_b_sq2cpsirgttcount",
analyticsigma_b_sqtext,
value = 0.1,
step = 0.001,
min = 0
)
)
),
#end of values that can be reset with the restore defaults button
actionButton(
"button",
"Estimate Power",
icon = icon("arrow-circle-right"),
width = '100%',
style = "color: #fff; background-color: #337ab7; border-color: #2e6da4"
),
actionButton('cancel', 'Cancel'),
checkboxInput("more", "Show advanced options", value = FALSE),
conditionalPanel(
"input.more == true",
bookmarkButton("Save App State"),
sliderInput(
"alpha",
"Significance level (alpha)",
value = 0.05,
min = 0.01,
max = 0.1,
step = 0.02
),
checkboxInput("verbose", "Show verbose results", value = FALSE),
checkboxInput("debug", "Show debug/diagnostics (advanced)", value = FALSE)
),
conditionalPanel(
"input.more == true && input.meth == 'Simulation'",
checkboxInput("timelimitOverride", "Allow unlimited calculation time", value = FALSE),
checkboxInput("lowPowerOverride", "Allow completion when power is < 0.5", value = FALSE),
checkboxInput(
"poorFitOverride",
"Allow completion when model fit is poor",
value = FALSE
),
textInput("optmethod", "Specify an optimization method", value = "NLOPT_LN_NELDERMEAD"),
numericInput(
"seed",
"Set the seed (for repeatability)",
value = NA,
step = 1
)
),
conditionalPanel(
"input.more == true",
actionButton(
"restoreDefault",
"Restore default parameters",
width = '100%',
style = "color: #fff; background-color: #337ab7; border-color: #2e6da4"
),
actionButton("reload", "Reset all", icon = icon("trash-alt"))
)
),
########################
# Tabs start
########################
mainPanel(tabsetPanel(
tabPanel(
"Results",
conditionalPanel(
"input.dismissMsg == false && input.dismissMsg2 == false && input.more == false",
wellPanel(
HTML(
"<p>Note: If the estimated calculation time is longer than 3 minutes,
the process will not run unless you override the time limit
under advanced options.</p>"
),
checkboxInput("dismissMsg", "dismiss this message", value = FALSE)
)
),
conditionalPanel(
"input.type == 'Stepped Wedge' && input.dist == 'Binary' &&
input.meth == 'Analytic' && input.stepscpaswbinary > 3 &&
input.dismissMsgCrossover == false",
wellPanel(
HTML(
"<p>Note: Including time effects or many crossover steps may
substantially increase calculation time. </p>"
),
checkboxInput("dismissMsgCrossover", "dismiss this message", value = FALSE)
)
),
verbatimTextOutput("CRTpower", placeholder = TRUE),
conditionalPanel("input.meth == 'Simulation'",),
#### DEBUG ACCESS PANEL START #####
conditionalPanel(
"input.debug == true",
textInput("debugSearch", "Search input", value = ""),
tableOutput("show_inputs")
)
#### DEBUG ACCESS PANEL END #####
),
tabPanel(
"Graphs",
conditionalPanel(
"input.dismissMsg == false && input.dismissMsg2 == false && input.more == false",
wellPanel(
HTML(
"<p>Note: If the estimated calculation time is longer than 3 minutes,
the process will not run unless you override the time limit
under advanced options.</p>"
),
checkboxInput("dismissMsg2", "dismiss this message", value = FALSE)
)
),
selectInput("axisname",
"Y-axis name",
choices = c("nclusters", "nsubjects")),
plotOutput("graphic", click = "click"),
tableOutput("dp"),
actionButton("cleargraph", "Clear Data", icon = icon("trash-alt")),
tags$style(type = 'text/css', "button#cleargraph { margin-top: 100px; }")
),
tabPanel(
"Parameters",
h5(
id = "helpnote",
"Note: This tab shows help for the
clusterPower functions. For help with this app, see
the Help tab."
),
shinyjs::hidden(
textInput("fxnName", "clusterPower function name", value = "cpa.normal")
),
wellPanel(
HTML(
"<p>This table shows the values that the Shiny app passes
to the R functions.</p>
<p>Note: for more advanced features, see the"
),
tags$a(
" clusterPower R package.",
href = "https://cran.r-project.org/web/packages/clusterPower/clusterPower.pdf",
target = "_blank"
)
),
fluidRow(
column(width = 6, tableOutput("tracker")),
column(width = 6)
),
downloadButton(
"downloadData",
"Download this table (.csv)",
icon = icon("file-download"),
style = "color: #fff; background-color: #337ab7; border-color: #2e6da4"
),
actionButton("cleargraph2", "Clear Data", icon = icon("trash-alt")),
tags$style(type = 'text/css', "button#cleargraph2 { margin-left: 250px; }")
),
tabPanel(
"Help",
h5(
id = "helpnote",
"Note: This tab shows help for this app.
For help with the clusterPower functions, see the Parameters tab."
),
tags$style(HTML("#helpnote{color: #337ab7;}")),
HTML(
"<p>To use the calculator, select the trial type, outcome distribution, and calculation method.
Then enter values for the quantities that appear below. When complete, select the ESTIMATE POWER button.</p>"
),
HTML("<h3>Getting started</h3>"),
HTML(
"<p>The clusterPower package is intended to perform power calculations for many of the most common
randomized controlled trial (RCT) designs. This app does not allow the user to access all of the
functions available in the clusterPower R package, such as calculating the numbers of clusters or
subjects needed to obtain a specific power, returning the raw simulated datasets, or viewing the
results of each model fit. For these functions, use clusterPower with the R console rather than
from within the applet.</p>
<p>The first and most important step for using this app is to choose the appropriate experimental
design and outcome distribution for your RCT. For more on this topic, consult the "
),
tags$a(
"clusterPower vignette.",
href = "https://cran.r-project.org/web/packages/clusterPower/vignettes/clusterpower.html",
target = "_blank"
),
HTML(
"To return to this page, click the back or reload button at the top of your browser window.</p>
<h4>Choosing a distribution</h4>
<p>After you have selected the RCT type using the pulldown menu, select the outcome distribution using
the next pulldown menu. The options here include normal, binary, and count distributions. Use normal
distribution when your measurement of interest is a numeric value. This can include measurements like
descriptive variables such as weights, lab results, density, etc. Choose binary distribution if your
outcome is a yes/no type of response metric, which would be found in studies with outcomes
represented by exactly two choices (sometimes qualitative), such as survived/deceased, uninfected/infected,
or participated/withdrew. Choose count when the outcome has more than two possible outcomes, such as
uninfected/infected/recovered/died.</p>
<h4>Choosing a method</h4>
<p>The user can choose the calculation method using the 'Method' dropdown menu. The choices are analytical
or simulation. Analytical methods have the advantage that calculations are sometimes faster than
simulated methods, but many make assumptions about variance or balance in design that may sacrifice
the accuracy of the power estimation. Simulated methods take longer to run but are more flexible because
they make fewer assumptions. Furthermore, analytical methods don't exist for all RCT types, meaning that in
those cases the simulation approach may be the user's only option. However, analytical and simulated
power estimation will likely produce similar results.</p>
<h4>Parameters</h4>
<p>Depending on the user's choices as outlined in the previous sections, different parameter entry
options will appear in the left-hand panel. All of these include the number of observations (or
subjects, depending on design requirements), and the number of clusters in each arm. Other options
include variables representing the expected outcomes for each arm, and those representing measures
of variablity among and within clusters. For simulated methods, the user can also supply a number
of simulations they would like to use for calculation.</p>
<p>Here the user may want to consult the 'Parameters' tab option, which displays the verbatim
parameter values that the app passes to the clusterPower backend as they are entered by the user.
This table also shows the internal argument names for each parameter.</p>
<h4>Advanced options</h4>
<p>The checkbox near the bottom of the left-hand panel opens the Advanced Options for the app. For
all analysis types, these options include options to bookmark the app state, restore the default
parameters, or reset (reload) the app. Users can also adjust the significance cutoff value alpha
using the slider. Typically this value is set to 0.05, which is the default for this app.</p>
<p>For the simulation methods, there are 3-4 additional controls. These include:</p>
<p>1) Time limit override: Simulation methods sometimes take a long time to complete. Each
operation will produce a time estimate based on a few model fits. If the estimate is longer than
2 minutes, the function will stop and return the estimated time to produce the desired fits.
Unchecking this option allows the app to run indefinitely, so users can remove the time limit
constraint if they choose.</p>
<p>2) Low power override: The simulated methods also have a built-in option to stop fitting
models if the calculated power is lower than 0.5. Generally, 0.8 is the ideal power target
for RCT power estimation, so estimations that are very low trigger the alogrithm to stop
fitting so that the user doesn't have to wait for the run to finish. However, the low power
error can be overridden by selecting the low power override option.</p>
<p>3) Poor fit override: When more than 25% of models fail to converge, the default app state
will stop the calculation with an error, again to prevent the user from waiting a long time
for uninformative results. Lack of convergence generally is an indication that the data is
not a good fit for the models. This is ideally addressed by adjusting the model parameters.
However, if the user wants to allow the procedure to run despite lack of convergence, the
poor fit override option will override this error. Use this option with caution, as models
lacking convergence may produce unreliable estimates.</p>
<p>4) Some simulated methods allow the user to specify an optimizer, which can sometimes
address convergence issues as an alternative to overriding the poor fit checks or excluding
non-convergent models.</p>
<h3>Results</h3>
<p>After selecting the desired parameters, submit the job by clicking the Estimate Power button
near the bottom of the screen. When complete, results will appear on the Results tab. Please
keep in mind that calculations may take up to 2 minutes to complete, unless the user has chosen
to override the time limit, in which case the wait time may be longer. Wait times vary depending
on the CRT design and complexity of the resulting model, whether the method is analytic or
simulation, and the amount of RAM available in the host computer. The Results tab shows the power
calculation resulting from the most recent clusterPower run, although consecutive runs are logged
unless the cached data is manually cleared, or the CRT type, method, or distribution is
changed by the user.</p>
<h3>Graphics</h3>
<p> As mentioned in the previous section, consecutive runs are logged
unless the cached data is manually cleared, or the CRT type, method, or distribution is
changed by the user. On the Graphics tab, the user can graph any of the user-selected
parameters against the resulting power estimate using the drop-down menu. Exact coordinates
for each point can be obtained by clicking on the point of interest. If the user would like
to clear the graph manually, the Clear Data button at the bottom of the panel will clear the
Results and Parameters tabs to their original state.</p>
<h3>Parameters</h3>
<p>On the parameters tab, consecutive clusterPower runs are logged until the data is cleared
or the user selects a different CRT type, method, or distribution. Parameters are shown
according to their argument names when passed to the clusterPower function. To learn more
about each parameter, select the 'Tell Me More About This' link near the top of the panel
to open a pop-up window showing the documentation for the clusterPower function in use.</p>
<h3>Note</h3>
<p>App created by Alexandria Sakrejda, Jon Moyer, and Ken Kleinman; support from NIGMS grant R01GM121370.
Please contact ken.kleinman@gmail.com with any feedback.</p>"
)
)
)) # Tabs end
)
)
######################################
# SERVER
######################################
server <- function(input, output, session) {
disable("cancel")
out1 <- reactiveValues(power = NULL)
logargs <- reactiveValues(tab = NULL)
# Reload the app
observeEvent(input$reload, {
session$reload()
})
# Restore default values
observeEvent(input$restoreDefault, {
shinyjs::reset("allValues")
})
#change text input to numeric
textToNum <- function(x) {
result <- as.numeric(unlist(strsplit(x, split = ",")))
return(result)
}
#make some helpful fxns to extract arg names
updateArgs <- function(fxnName) {
argMatchResult <-
c(
clusterPower::argMatch(fxnName, justNames = TRUE),
"lowPowerOverride",
"poorFitOverride",
"timelimitOverride",
"power",
"seed",
"optmethod"
)
argNames <-
c("nsubjects",
"nclusters",
"alpha",
dplyr::intersect(argMatchResult,
names(formals(fxnName))))
arghelper <- function(argname) {
x <- paste0("input$", argname)
return(x)
}
holder <- list()
for (i in 1:length(argNames)) {
holder[[i]] <- arghelper(argNames[i])
}
names(holder) <- argNames
return(holder)
}
printresult <- function(fxnName) {
x <- rlang::exec(fxnName, !!!updateArgs(fxnName))
return(x)
}
#end of fxns to extract argument names
# update help documentation and params table when function is selected
observe({
if (input$type == 'Parallel' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.count")
shinyjs::show(id = "button")
}
if (input$type == 'Parallel' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.count")
shinyjs::show(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.ma.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.ma.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.ma.binary")
shinyjs::hide(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.ma.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.ma.count")
shinyjs::hide(id = "button")
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.ma.count")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.did.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.did.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.did.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.did.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.did.count")
shinyjs::hide(id = "button")
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.did.count")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.sw.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.sw.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.sw.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.sw.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.sw.count")
shinyjs::show(id = "button")
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.sw.count")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.irgtt.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.irgtt.normal")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.irgtt.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.irgtt.binary")
shinyjs::show(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Count' && input$meth == 'Analytic') {
updateTextInput(session, "fxnName", value = "cpa.irgtt.count")
shinyjs::hide(id = "button")
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Count' && input$meth == 'Simulation') {
updateTextInput(session, "fxnName", value = "cps.irgtt.count")
shinyjs::show(id = "button")
}
}) # end update help documentation and params table when function is selected
# call the clusterPower functions
observeEvent(input$button, {
disable("button")
enable("cancel")
prog <- Progress$new(session)
prog$set(message = "Analysis in progress",
value = NULL)
isolate({
q <- reactiveValuesToList(input)
})
if (input$type == 'Parallel' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <-
cpa.normal(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpanormal,
nsubjects = q$nsubjectscpanormal,
sigma_sq = q$sigma_sqcpanormal,
sigma_b_sq = q$sigma_b_sqcpanormal,
CV = q$CVcpanormal,
d = q$dcpanormal,
ICC = q$ICCcpanormal,
vart = q$vartcpanormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.normal(
nsim = q$nsimcpsnormal,
nclusters = c(q$nclusters1cpsnormal, q$nclusters2cpsnormal),
nsubjects = textToNum(c(
q$nsubjects1cpsnormal, q$nsubjects2cpsnormal
)),
mu = q$mucpsnormal,
mu2 = q$mu2cpsnormal,
ICC = q$ICCcpsnormal,
sigma_sq = q$sigma_sqcpsnormal,
sigma_b_sq = q$sigma_b_sqcpsnormal,
ICC2 = q$ICC2cpsnormal,
sigma_sq2 = q$sigma_sq2cpsnormal,
sigma_b_sq2 = q$sigma_b_sq2cpsnormal,
alpha = q$alpha,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
timelimitOverride = q$timelimitOverride,
lowPowerOverride = q$lowPowerOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.binary(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpabinary,
nsubjects = q$nsubjectscpabinary,
CV = q$CVcpabinary,
p1 = q$p1cpabinary,
p2 = q$p2cpabinary,
ICC = q$ICCcpabinary,
pooled = q$pooledcpabinary,
p1inc = q$p1inccpabinary,
tdist = q$tdistcpabinary
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.binary(
nsim = q$nsimcpsbinary,
nsubjects = textToNum(c(
q$nsubjects1cpsbinary, q$nsubjects2cpsbinary
)),
nclusters = c(q$nclusters1cpsbinary, q$nclusters2cpsbinary),
p1 = q$p1cpsbinary,
p2 = q$p2cpsbinary,
sigma_b_sq = q$sigma_b_sqcpsbinary,
sigma_b_sq2 = q$sigma_b_sq2cpsbinary,
alpha = q$alpha,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.count(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpacount,
nsubjects = q$nsubjectscpacount,
r1 = q$r1cpacount,
r2 = q$r2cpacount,
CVB = q$CVBcpacount,
r1inc = q$r1inccpacount
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Parallel' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.count(
nsim = q$nsimcpscount,
nsubjects = textToNum(c(
q$nsubjects1cpscount, q$nsubjects2cpscount
)),
nclusters = c(q$nclusters1cpscount, q$nclusters2cpscount),
c1 = q$c1cpscount,
c2 = q$c2cpscount,
sigma_b_sq = q$sigma_b_sqcpscount,
sigma_b_sq2 = q$sigma_b_sq2cpscount,
alpha = q$alpha,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.ma.normal(
alpha = q$alpha,
power = q$power,
narms = q$narmscpamanormal,
nclusters = q$nclusterscpamanormal,
nsubjects = q$nsubjectscpamanormal,
vara = q$varacpamanormal,
varc = q$varccpamanormal,
vare = q$varecpamanormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.ma.normal(
nsim = q$nsimcpsmanormal,
nsubjects = textToNum(q$nsubjectscpsmanormal),
narms = q$narmscpsmanormal,
nclusters = textToNum(q$nclusterscpsmanormal),
means = textToNum(q$meanscpsmanormal),
sigma_sq = textToNum(q$sigma_sqcpsmanormal),
sigma_b_sq = textToNum(q$sigma_b_sqcpsmanormal),
alpha = q$alpha,
ICC = textToNum(q$ICCcpsmanormal),
multi_p_method = q$multi_p_methodcpsmanormal,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
tdist = q$tdistcpsmanormal,
optmethod = q$optmethod,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
answer <<- future({
val <- cpa.ma.binary()
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.ma.binary(
nsim = q$nsimcpsmabinary,
nsubjects = textToNum(q$nsubjectscpsmabinary),
narms = q$narmscpsmabinary,
nclusters = textToNum(q$nclusterscpsmabinary),
probs = textToNum(q$probscpsmabinary),
sigma_b_sq = textToNum(q$sigma_b_sqcpsmabinary),
alpha = q$alpha,
multi_p_method = q$multi_p_methodcpsmabinary,
seed = q$seed,
tdist = q$tdistcpsmabinary,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
answer <<- future({
val <- cpa.ma.count()
return(val)
}, seed = TRUE)
}
if (input$type == 'Multi-Arm' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.ma.count(
nsim = q$nsimcpsmacount,
nsubjects = textToNum(q$nsubjectscpsmacount),
narms = q$narmscpsmacount,
nclusters = textToNum(q$nclusterscpsmacount),
counts = textToNum(q$countscpsmacount),
sigma_b_sq = textToNum(q$sigma_b_sqcpsmacount),
alpha = q$alpha,
multi_p_method = q$multi_p_methodcpsmacount,
seed = q$seed,
tdist = q$tdistcpsmacount,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.did.normal(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpadidnormal,
nsubjects = q$nsubjectscpadidnormal,
d = q$dcpadidnormal,
ICC = q$ICCcpadidnormal,
rho_c = q$rho_ccpadidnormal,
rho_s = q$rho_scpadidnormal,
vart = q$vartcpadidnormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.did.normal(
nsim = q$nsimcpsdidnormal,
nsubjects = textToNum(
c(q$nsubjects1cpsdidnormal, q$nsubjects2cpsdidnormal)
),
nclusters = c(q$nclusters1cpsdidnormal, q$nclusters2cpsdidnormal),
delta_mu = q$delta_mucpsdidnormal,
delta_mu2 = q$delta_mu2cpsdidnormal,
sigma_sq = q$sigma_sqcpsdidnormal,
sigma_b_sq0 = c(
q$sigma_b_sq01cpsdidnormal,
q$sigma_b_sq02cpsdidnormal
),
sigma_b_sq1 = c(
q$sigma_b_sq11cpsdidnormal,
q$sigma_b_sq12cpsdidnormal
),
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.did.binary(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpadidbinary,
nsubjects = q$nsubjectscpadidbinary,
p = q$pcpadidbinary,
d = q$dcpadidbinary,
ICC = q$ICCcpadidbinary,
rho_c = q$rho_ccpadidbinary,
rho_s = q$rho_scpadidbinary
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.did.binary(
nsim = q$nsimcpsdidbinary,
nsubjects = textToNum(
c(q$nsubjects1cpsdidbinary, q$nsubjects2cpsdidbinary)
),
nclusters = c(q$nclusters1cpsdidbinary, q$nclusters2cpsdidbinary),
p1t0 = q$p1t0cpsdidbinary,
p2t0 = q$p2t0cpsdidbinary,
p1t1 = q$p1t1cpsdidbinary,
p2t1 = q$p2t1cpsdidbinary,
sigma_b_sq0 = c(
q$sigma_b_sq01cpsdidbinary,
q$sigma_b_sq02cpsdidbinary
),
sigma_b_sq1 = c(
q$sigma_b_sq11cpsdidbinary,
q$sigma_b_sq12cpsdidbinary
),
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
answer <<- future({
val <- cpa.did.count()
return(val)
}, seed = TRUE)
}
if (input$type == 'Difference-in-Difference' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.did.count(
nsim = q$nsimcpsdidcount,
nsubjects = textToNum(c(
q$nsubjects1cpsdidcount, q$nsubjects2cpsdidcount
)),
nclusters = c(q$nclusters1cpsdidcount, q$nclusters2cpsdidcount),
c1t0 = q$c1t0cpsdidcount,
c2t0 = q$c2t0cpsdidcount,
c1t1 = q$c1t1cpsdidcount,
c2t1 = q$c2t1cpsdidcount,
sigma_b_sq0 = c(
q$sigma_b_sq01cpsdidcount,
q$sigma_b_sq02cpsdidcount
),
sigma_b_sq1 = c(
q$sigma_b_sq11cpsdidcount,
q$sigma_b_sq12cpsdidcount
),
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.sw.normal(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpaswnormal,
nsubjects = q$nsubjectscpaswnormal,
ntimes = q$ntimescpaswnormal,
d = q$dcpaswnormal,
ICC = q$ICCcpaswnormal,
rho_c = q$rho_ccpaswnormal,
rho_s = q$rho_scpaswnormal,
vart = q$vartcpaswnormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.sw.normal(
alpha = q$alpha,
nsim = q$nsimcpsswnormal,
nclusters = q$nclusterscpsswnormal,
nsubjects = textToNum(q$nsubjectscpsswnormal),
steps = q$stepscpsswnormal,
mu0 = q$mu0cpsswnormal,
mu1 = q$mu1cpsswnormal,
sigma_sq = q$sigma_sqcpsswnormal,
sigma_b_sq = q$sigma_b_sqcpsswnormal,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.sw.binary(
alpha = q$alpha,
nclusters = q$nclusterscpaswbinary,
steps = q$stepscpaswbinary,
nsubjects = q$nsubjectscpaswbinary,
timeEffect = q$timeEffectcpaswbinary,
ICC = q$ICCcpaswbinary,
p1 = q$p1cpaswbinary,
p0 = q$p0cpaswbinary
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.sw.binary(
nsim = q$nsimcpsswbinary,
nsubjects = textToNum(q$nsubjectscpsswbinary),
nclusters = q$nclusterscpsswbinary,
p0 = q$p0cpsswbinary,
p1 = q$p1cpsswbinary,
steps = q$stepscpsswbinary,
sigma_b_sq = q$sigma_b_sqcpsswbinary,
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.sw.count(
lambda1 = q$lambda1cpaswcount,
RR = q$RRcpaswcount,
nclusters = q$nclusterscpaswcount,
steps = q$stepscpaswcount,
nsubjects = q$nsubjectscpaswcount,
ICC = q$ICCcpaswcount,
alpha = q$alpha
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Stepped Wedge' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.sw.count(
nsim = q$nsimcpsswcount,
nsubjects = textToNum(q$nsubjectscpsswcount),
nclusters = q$nclusterscpsswcount,
c0 = q$c0cpsswcount,
c1 = q$c1cpsswcount,
steps = q$stepscpsswcount,
sigma_b_sq = q$sigma_b_sqcpsswcount,
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Normal' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.irgtt.normal(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpairgttnormal,
nsubjects = q$nsubjectscpairgttnormal,
ncontrols = q$ncontrolscpairgttnormal,
d = q$dcpairgttnormal,
varu = q$varucpairgttnormal,
varei = q$vareicpairgttnormal,
varr = q$varrcpairgttnormal
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Normal' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.irgtt.normal(
nsim = q$nsimcpsirgttnormal,
nsubjects = textToNum(
c(
q$nsubjectscpsirgttnormal,
q$nsubjects2cpsirgttnormal
)
),
nclusters = q$nclusterscpsirgttnormal,
mu = q$mucpsirgttnormal,
mu2 = q$mu2cpsirgttnormal,
sigma_sq = q$sigma_sqcpsirgttnormal,
ICC2 = q$ICC2cpsirgttnormal,
sigma_sq2 = q$sigma_sq2cpsirgttnormal,
sigma_b_sq2 = q$sigma_b_sq2cpsirgttnormal,
alpha = q$alpha,
seed = q$seed,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Binary' && input$meth == 'Analytic') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cpa.irgtt.binary(
alpha = q$alpha,
power = q$power,
nclusters = q$nclusterscpairgttbinary,
nsubjects = q$nsubjectscpairgttbinary,
ncontrols = q$ncontrolscpairgttbinary,
ICC = q$ICCcpairgttbinary,
p2 = q$p2cpairgttbinary,
p1 = q$p1cpairgttbinary,
decrease = q$decreasecpairgttbinary
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Binary' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.irgtt.binary(
nsim = q$nsimcpsirgttbinary,
nsubjects = textToNum(
c(
q$nsubjectscpsirgttbinary,
q$nsubjects2cpsirgttbinary
)
),
nclusters = q$nclusterscpsirgttbinary,
p1 = q$p1cpsirgttbinary,
p2 = q$p2cpsirgttbinary,
sigma_b_sq2 = q$sigma_b_sq2cpsirgttbinary,
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Count' && input$meth == 'Analytic') {
out1$power <- NULL
answer <<- future({
val <- cpa.irgtt.count()
return(val)
}, seed = TRUE)
}
if (input$type == 'Individually-Randomized Group' &&
input$dist == 'Count' && input$meth == 'Simulation') {
out1$power <- NULL
out1$power$power <- "Calculating..."
answer <<- future({
val <- cps.irgtt.count(
nsim = q$nsimcpsirgttcount,
nsubjects = textToNum(
c(q$nsubjectscpsirgttcount, q$nsubjects2cpsirgttcount)
),
nclusters = q$nclusterscpsirgttcount,
c1 = q$c1cpsirgttcount,
c2 = q$c2cpsirgttcount,
sigma_b_sq2 = q$sigma_b_sq2cpsirgttcount,
alpha = q$alpha,
poorFitOverride = q$poorFitOverride,
lowPowerOverride = q$lowPowerOverride,
timelimitOverride = q$timelimitOverride,
seed = q$seed
)
return(val)
}, seed = TRUE)
}
then(
answer,
onFulfilled = function(value) {
for (i in names(value)) {
out1$power[[i]] <- value[[i]]
}
out1 <<- out1
},
onRejected = function(error) {
out1$power$power <- paste0("ERROR: ", error$message)
}
)
finally(answer, function() {
prog$close()
disable("cancel")
enable("button")
})
return(NULL)
}, ignoreInit = TRUE) # end call the clusterPower functions
##################
# cancel button
observeEvent(input$cancel, {
async_pid <- answer$process$get_pid()
print(paste("Killing PID:", async_pid))
answer$process$kill()
out1$power <- NULL
disable("cancel")
enable("run1")
}, ignoreInit = TRUE)
#########################################
#### DEBUG ACCESS PANEL #####
#########################################
AllInputs <- reactive({
x <- reactiveValuesToList(input)
holderDebug <- NULL
if (sum(grepl("click", names(x))) == 1) {
x$click <- NULL
}
holderDebug <- data.frame(
names = names(isolate(x)),
values = unlist(isolate(x), use.names = FALSE),
mode = unlist(lapply(isolate(x), mode))
)
holderDebug <- holderDebug[order(holderDebug$names), ]
if (input$debugSearch != "") {
holderDebug <- holderDebug[grepl(x$debugSearch, holderDebug$names), ]
}
return(holderDebug)
})
output$show_inputs <- renderTable({
AllInputs()
})
######################################### END DEBUG table
#create the graphing table
observeEvent(req(out1$power$power), {
if (is.character(isolate(out1$power$power)) == FALSE) {
x <- reactiveValuesToList(input)
holder <- NULL
if (sum(grepl("click", names(x))) == 1) {
x$click <- NULL
}
holder <- data.frame(
argument = names(isolate(x)),
values = unlist(isolate(x), use.names = FALSE)
)
specialnames <-
dplyr::filter(holder, grepl(gsub("\\.", "", input$fxnName), argument))
specialnames$argument <-
gsub(gsub("\\.", "", input$fxnName),
"",
specialnames$argument)
specialnames <- dplyr::arrange(specialnames, argument)
if (x$meth == "Analytic") {
tab <-
rbind(specialnames,
c("alpha", isolate(input$alpha)),
c("power", round(out1$power$power, 3)))
}
if (x$meth == "Simulation") {
tab <-
rbind(
specialnames,
c("alpha", isolate(input$alpha)),
c("upper CI", round(out1$power$power$Upper.95.CI, 3)),
c("power", round(out1$power$power$Power, 3)),
c("lower CI", round(out1$power$power$Lower.95.CI, 3))
)
}
if (is.null(logargs$tab)) {
logargs$tab <- tab
} else {
tab <- dplyr::select(tab, values)
tab <- cbind.data.frame(logargs$tab, tab)
logargs$tab <- data.frame(tab, check.names = TRUE)
}
} else {
# if logargs$tab is an error, ignore it
if (!is.null(logargs$tab)) {
tab <- cbind.data.frame(logargs$tab)
logargs$tab <- data.frame(tab, check.names = TRUE)
}
}
}) # END create the graphing table
#clear the data log under certain circumstances
observeEvent(input$cleargraph, {
logargs$tab <- NULL
})
observeEvent(input$cleargraph2, {
logargs$tab <- NULL
})
observeEvent(input$fxnName, {
logargs$tab <- NULL
})
# END clear the data log under certain circumstances
# make the graph
#update the axis choices
observeEvent(input$fxnName, {
x <- reactiveValuesToList(input)
holder <- names(x)
specialnames <-
grep(gsub("\\.", "", isolate(input$fxnName)), holder, value = TRUE)
specialnames <-
gsub(gsub("\\.", "", isolate(input$fxnName)), "", specialnames)
specialnames <-
specialnames[grepl("nsim", specialnames) == FALSE]
args_ <- c("alpha", specialnames)
updateSelectInput(session, "axisname",
choices = args_)
})
plot_this <-
eventReactive(list(input$axisname, req(logargs$tab)), {
data <- data.frame(isolate(logargs$tab), check.names = TRUE)
data$values <- as.numeric(data$values)
data$argument <- as.factor(data$argument)
var <- isolate(input$axisname)
data <-
dplyr::filter(data, argument == !!var |
argument == "power")
# first remember the names
n <- data$argument
# transpose all but the first column (name)
data <- data.frame(t(data[, -1]))
colnames(data) <- n
data[, 1:2] <- apply(data[, 1:2], 2,
function(x)
as.numeric(as.character(x)))
return(data)
})
dpfun <- function(x) {
var <- isolate(input$axisname)
data <- plot_this()
fun <- function(x) {
x <- enquo(x)
sol <-
ggplot(data, aes(x = !!x, y = power)) +
geom_point(aes(colour = "fff"), size = 2.5) +
theme_minimal() + theme(legend.position = "none")
return(sol)
}
power_plot <- suppressWarnings(fun(get(var)))
if (nrow(data) > 1) {
power_plot <-
power_plot + geom_line(aes(colour = "fff"), size = 1.25) + xlab(var)
} else {
power_plot <- power_plot + xlab(var)
}
return(power_plot)
}
output$graphic <- renderPlot({
dpfun()
}, res = 96)
output$dp <- renderTable({
q <- plot_this()
nearPoints(q, isolate(input$click), xvar = isolate(input$axisname))
})
# create reactive input data table
output$tracker <-
renderTable(logargs$tab)
# Downloadable csv of reactive input data table
output$downloadData <- downloadHandler(
filename = function() {
paste("clusterPower_", isolate(input$fxnName), ".csv", sep = "")
},
content = function(file) {
write.csv(logargs$tab, file, row.names = FALSE)
}
)
# main power output
resultdisplay <- reactive({
q <- reactiveValuesToList(out1)
if (input$verbose == FALSE)
return(q$power$power)
else
return(q$power)
})
# present the output verbose/not verbose
output$CRTpower <- renderPrint(resultdisplay())
} #end of server fxn
# Run the application
shinyApp(ui = ui, server = server)
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.