Evaluating and Learning Optimal Dynamic Treatment Regimes under Truncation by Death
This addresses a critical challenge in healthcare for optimizing personalized treatments when patients may die during care, though it appears incremental as it builds on existing principal stratification and DTR frameworks.
The paper tackles the problem of evaluating dynamic treatment regimes when patients die during treatment (truncation by death), which makes traditional methods inapplicable. It introduces a principal stratification-based method with a semiparametrically efficient, multiply robust estimator, validated empirically and on electronic health records for personalized treatment optimization.
Truncation by death, a prevalent challenge in critical care, renders traditional dynamic treatment regime (DTR) evaluation inapplicable due to ill-defined potential outcomes. We introduce a principal stratification-based method, focusing on the always-survivor value function. We derive a semiparametrically efficient, multiply robust estimator for multi-stage DTRs, demonstrating its robustness and efficiency. Empirical validation and an application to electronic health records showcase its utility for personalized treatment optimization.