This repository includes the code and sample dataset required to make predictions and fit a treatment selection model augmented with a Dirichlet process mixture model (DPMM). The DPMM enables the ability to fit models and make predictions in the presence of missing values (assuming these are missing completely at random or missing at random).
Paper: Pedro Cardoso, John M. Dennis, Jack Bowden, Beverley M. Shields, Trevelyan J. McKinley the MASTERMIND Consortium. Dirichlet process mixture models to impute missing predictor data in counterfactual prediction models: an application to predict optimal type 2 diabetes therapy. BMC Medical Informatics and Decision Making 24, 12 (2024), doi: <https://doi.org/10.1186/s12911-023-02400-3>.
Package: DPMM is an R package providing a library of model fitting functions, diagnostics tools for the fitted model and plotting functions. https://github.com/PM-Cardoso/DPMM