Primary Submission Category: Machine Learning and Causal Inference
Causal Inference and Adaptive Design for Evaluating Effectiveness of Medical Tests and Devices
Authors: Wenxin Zhang, Rachael Phillips, Gene Pennello, Mark van der Laan,
Presenting Author: Wenxin Zhang*
Diagnostic medical devices, i.e., tests, play a critical role in detecting and monitoring diseases. However, unlike treatments, the effects of test results on health outcomes are indirect through their downstream influence on treatment decisions. This disconnection poses a challenge in evaluating the true effectiveness of tests. In this work, we propose causal estimands to evaluate effectiveness of medical tests from their explanatory and/or pragmatic utility in medical care, estimating them by the Targeted Maximum Likelihood Estimation (TMLE) method. We further propose an adaptive experiment design to better evaluate medical tests and devices. This framework is generally applicable to evaluate the effectiveness of any object (e.g. AI prediction score) whose effect on the primary outcome of interest is mediated by the consequent treatment decision. The performance of our estimators and designs is demonstrated through simulation studies.