estimate_evppi: Function to estimate the expected value of partial perfect...

In sheejamk/packEVPI: Estimate Expected Value of Perfect Information

Function to estimate the expected value of partial perfect information This is the outer loop of the two stage Monte Carlo process

Description

Function to estimate the expected value of partial perfect information This is the outer loop of the two stage Monte Carlo process

Usage

estimate_evppi(
  params_interest,
  param_file,
  colnames_paramdistr,
  list_markov,
  threshold,
  outer_iterations,
  inner_iterations,
  comparator = NULL
)

Arguments

params_interest	the parameters of interest
param_file	all parameters required to run the model,provided with name of parameter, distribution and parameters that define the probability distribution
colnames_paramdistr	column names where parameter distribution and values are defined
list_markov	list of markov models to estimate the NMB
threshold	threshold values of WTP
outer_iterations	number of iterations for outer loop
inner_iterations	number of iterations for inner loop
comparator	optional parameter if need ICER as the output

Value

result result after evppi calculation, including three forms of evppi

Source

https://www.sciencedirect.com/science/article/pii/S1098301510605888

Examples

 colnames_paramdistr  <- c("Param1_name", "Param1_value", "Param2_name",
 "Param2_value")
 inner_iterations <- 5
 outer_iterations <- 2
 threshold <- 2000
 param_interest <- "tp_well_dis_co"
 value_param_interest <- 0.2
well <- packDAMipd::health_state("well", cost = "cost_well_co", utility = 1)
disabled <- packDAMipd::health_state("disabled", cost = "cost_dis_co",
utility = "utility_dis_co")
dead <- packDAMipd::health_state("dead", cost = 0, utility = 0)
tmat <- rbind(c(1, 2, 3), c(NA, 4, 5), c(NA, NA, 6))
colnames(tmat) <- rownames(tmat) <- c("well", "disabled", "dead")
tm <- packDAMipd::populate_transition_matrix(3, tmat,
c("tp_well_well_co","tp_well_dis_co","tp_well_dead", "tp_dis_dis_co",
"tp_dis_dead", "tp_dead_dead"),colnames(tmat))
health_states <- packDAMipd::combine_state(well, disabled, dead)
this.strategy <- packDAMipd::strategy(tm, health_states, "control")
param_file <- system.file("extdata", "table_param.csv",
package = "packEVPI")
param_list <- packDAMipd::define_parameters(
tp_well_dis_co = packDAMipd::get_parameter_read("tp_well_dis_co",
param_file),
tp_well_dis_in =  packDAMipd::get_parameter_read("tp_well_dis_in",
param_file),
tp_well_dead =  packDAMipd::get_parameter_read("tp_well_dead", param_file),
tp_dis_dead =  packDAMipd::get_parameter_read("tp_dis_dead", param_file),
 tp_dead_dead =  1,
 cost_well_co =  packDAMipd::get_parameter_read("cost_well_co", param_file),
 cost_well_in =  packDAMipd::get_parameter_read("cost_well_in", param_file),
 cost_dis_co =  packDAMipd::get_parameter_read("cost_dis_co", param_file),
 cost_dis_in =  packDAMipd::get_parameter_read("cost_dis_in", param_file),
 utility_dis_co =  packDAMipd::get_parameter_read("utility_dis_co",
 param_file),
 utility_dis_in =  packDAMipd::get_parameter_read("utility_dis_in",
 param_file),
 tp_well_well_co = "1-(tp_well_dis_co + tp_well_dead)",
 tp_well_well_in = "1-(tp_well_dis_in + tp_well_dead)",
 tp_dis_dis_co = "1-( tp_dis_dead)",
 tp_dis_dis_in = "1-( tp_dis_dead)")
 this_markov <- packDAMipd::markov_model(this.strategy, 24, c(1000, 0, 0),
 discount = c(0, 0),method = "half cycle correction", param_list)
 well <- packDAMipd::health_state("well", cost = "cost_well_in", utility = 1)
 disabled <- packDAMipd::health_state("disabled", cost = "cost_dis_in",
 utility = "utility_dis_in")
 dead <- packDAMipd::health_state("dead", cost = 0, utility = 0)
 tmat <- rbind(c(1, 2, 3), c(NA, 4, 5), c(NA, NA, 6))
 colnames(tmat) <- rownames(tmat) <- c("well", "disabled", "dead")
 tm <- packDAMipd::populate_transition_matrix(3, tmat, c("tp_well_well_in",
 "tp_well_dis_in", "tp_well_dead", "tp_dis_dis_in","tp_dis_dead",
 "tp_dead_dead"), colnames(tmat))
 health_states <- packDAMipd::combine_state(well, disabled, dead)
 this.strategy <- packDAMipd::strategy(tm, health_states, "intervention")
 sec_markov <- packDAMipd::markov_model(this.strategy, 24, c(1000, 0, 0),
 discount = c(0, 0), method = "half cycle correction", param_list)
 list_markov <- packDAMipd::combine_markov(list(this_markov, sec_markov))
 param_interest <- "tp_well_dis_co"
 res <- estimate_evppi(param_interest, param_file,
 colnames_paramdistr, list_markov, threshold, outer_iterations,
 inner_iterations)

sheejamk/packEVPI documentation built on April 7, 2023, 8:48 a.m.