In DARTH-git/darthpack: The Decision Analysis in R for Technologies in Health (DARTH) coding framework

library(knitr)
# https://haozhu233.github.io/kableExtra/awesome_table_in_html.html
library(kableExtra)
library(dplyr)
library(reshape2)
library(ggplot2)
library(darthpack)

Validation {#validation}

In this forth component, we check the internal validity of our Sick-Sicker model before we move on to the analysis components. To internally validate the Sick-Sicker model, we compare the model-predicted output evaluated at posterior parameters against the calibration targets. This is all done in the 04_validation.R script in the analysis folder.

In section 04.2 Compute model-predicted outputs, we compute the model-predicted outputs for each sample of posterior distribution as well as for the MAP estimate. We then use the function data_summary to summarize the model-predicted posterior outputs into different summary statistics.

print.function(data_summary)

This function is informed by three arguments, data, varname and groupnames.

The computation of the model-predicted outputs using the MAP estimate is done by inserting the v_calib_post_map data into the previously described calibration_out function. This function creates a list including the estimated values for survival, prevalence and the proportion of sicker individuals at cycles 10, 20 and 30.

In sections 04.6 Internal validation: Model-predicted outputs vs. targets, we check the internal validation by plotting the model-predicted outputs against the calibration targets (Figures \@ref(fig:04-surv)-\@ref(fig:04-proportion)). The generated plots are saved as .png files in the figs folder. These files can be used in reports without the need of re-running the code.

knitr::include_graphics("../figs/04_posterior_vs_targets_survival.png")

knitr::include_graphics("../figs/04_posterior_vs_targets_prevalence.png")

knitr::include_graphics("../figs/04_posterior_vs_targets_proportion_sicker.png")

DARTH-git/darthpack documentation built on March 10, 2024, 3:31 p.m.