Comparing predictive models, for their performance, cost and participant time needed

The goal of this project is to assess and compare the performance of different predictive model in term of discriminative power, calibration, cost and participant time thanks to the dataset generated by the Trialnet network.

The project is split in three parts; analysis, article and model parameters.

Model fitting and model assement

The main script to generate all the model and assess their performances can be found in main_load.R

Article preparation

The main file to generate the figures are in the directory R/article and in article/tables_quarto

Predictive model objects

the Predictive model R objects can be found in the pareto_front_models folder. The model coefficient are in the file model_coefficients.xlsx.