scripts/run_analysis_S3.R

In ICON-in-R/NMA: Easy Network Meta-Analysis

# N Green, UCL 7-12-2020 --------------------------------------------------

library(dplyr)
library(purrr)


## settings

bugs_params <-
  list(
    PROG = "openBugs",
    N.BURNIN = 10,#00,
    N.SIMS = 150,#0,
    N.CHAINS = 2,
    N.THIN = 1,
    PAUSE = TRUE)

hyperparams <-
  list(mu_beta = 0,
       prec_beta = 1.0E-6,
       mu_alpha = 0,
       prec_alpha = 1.0E-6)


## run analysis

analyses_params <-
  read.csv(
    here::here("raw_data", "AnalysisList.csv"),
    as.is = TRUE,
    na.strings = c("NR", "NA")) %>% 
  filter(Endpoint_type == "Surv") %>% 
  dplyr::rename(name = Analysis_name,
                type = Analysis_Type)


analysis <- analyses_params[1, ]

# fixed effects RANDOM=FALSE, random effects RANDOM=TRUE
RANDOM <- analysis$Model_effects == "RE"

REFTX <- analysis$REFTX

# indicator for availability of binary endpoint data
is_bin <- analysis$BinData == "YES"

# indicator for availability of median endpoint data
is_med <- analysis$MedData == "YES"


## read in datasets

file_name <- paste0(here::here("raw_data"), "/survdata_", analysis$Endpoint, "_")

subData <-
  read.csv(paste0(file_name, analysis$type, ".csv"),
           header = TRUE,
           as.is = TRUE)

survDataBin <-
  if (is_bin) {
    read.csv(paste0(file_name, "bin.csv"),
             header = TRUE,
             as.is = TRUE)
  } else {NA}

survDataMed <-
  if (is_med) {
    read.csv(paste0(file_name, "med.csv"),
             header = TRUE,
             as.is = TRUE) %>% 
      mutate(medR = floor(medR))
  } else {NA}


# nma_dat <- load_input_data(is_med, is_bin)

## build model
nma_model <-
  new_NMA(subData = subData,
          survDataMed = survDataMed,
          survDataBin = survDataBin,
          bugs_params = bugs_params,
          is_random = RANDOM,
          hyperparams = hyperparams,
          refTx = REFTX ,
          effectParam = "beta",
          modelParams = "totresdev",
          label = analysis$name,
          endpoint = analysis$Endpoint)

##TODO: why doesnt it work without attaching sna?
library(sna)
plotNetwork(nma_model)


## create output
nma_res <- NMA_run(nma_model)

diagnostics(nma_res)
nma_outputs(nma_res)

## reconfigure model
nma_model2 <-
  NMA_update(nma_model,
             is_random = TRUE)

nma_res2 <- NMA_run(nma_model2)

diagnostics(nma_res2, save = TRUE)
nma_outputs(nma_res2, save = TRUE)