A1840
Title: When-if decision-making using synthetic survival control
Authors: Jessy Xinyi Han - Broad Institute / MIT (United States) [presenting]
Abstract: Understanding the impact of decisions on when a target event occurs, not just whether it occurs, is central to many fields, including patient Survival in healthcare, criminal recidivism in policy evaluation, and customer churning in business. A when-if decision-making framework integrates causal inference and Survival analysis to support decisions with observational data where the timing of the target event is the key quantity of interest and the data are often sparse, censored, or confounded. The motivation concerns evaluating the efficacy of different therapies for T-Cell Lymphoma across a heterogeneous patient population. Synthetic Survival Control (SSC) is a new method for estimating counterfactual hazard trajectories in panel data with censoring and unobserved confounding. The method extends the traditional panel data literature by utilizing a causal Survival panel framework with an underlying low-rank structure that naturally arises under classical parametric Survival models. Within this framework, identification of the causal estimand and finite-sample guarantees for SSC are established. Full validation of the proposed method is provided through application to T-Cell Lymphoma treatment evaluation in collaboration with clinicians at Massachusetts General Hospital.
