Deconfounded Warm-Start Thompson Sampling with Applications to Precision Medicine
This work addresses the challenge of making clinical trials more efficient and personalized by integrating observational data, though it appears incremental as it builds on existing Thompson Sampling and deconfounding techniques.
The paper tackled the problem of underutilizing observational data in clinical trials due to confounding and biases, proposing Deconfounded Warm-Start Thompson Sampling (DWTS) to improve trial efficiency by leveraging offline data, which achieved lower cumulative regret than standard methods in synthetic and real cardiovascular risk evaluations.
Randomized clinical trials often require large patient cohorts before drawing definitive conclusions, yet abundant observational data from parallel studies remains underutilized due to confounding and hidden biases. To bridge this gap, we propose Deconfounded Warm-Start Thompson Sampling (DWTS), a practical approach that leverages a Doubly Debiased LASSO (DDL) procedure to identify a sparse set of reliable measured covariates and combines them with key hidden covariates to form a reduced context. By initializing Thompson Sampling (LinTS) priors with DDL-estimated means and variances on these measured features -- while keeping uninformative priors on hidden features -- DWTS effectively harnesses confounded observational data to kick-start adaptive clinical trials. Evaluated on both a purely synthetic environment and a virtual environment created using real cardiovascular risk dataset, DWTS consistently achieves lower cumulative regret than standard LinTS, showing how offline causal insights from observational data can improve trial efficiency and support more personalized treatment decisions.