Learning Personalized Treatment Strategies with Predictive and Prognostic Covariates in Adaptive Clinical Trials

Abstract

Biomedical research may uncover insights regarding the interaction of the treatments of a disease with patient covariates. We show how to use such insights to improve the efficiency of adaptive clinical trials for precision medicine by extending ideas from optimal Bayesian learning. We present a model for response-adaptive multi-arm clinical trials that leverages knowledge about the predictive-prognostic covariate structure to accelerate the learning of personalized treatment strategies that obtain the best expected outcomes for posttrial patients. Our base model is a contextual linear bandit for best-arm identification, and outcomes may be observed with delay.

We characterize the optimal policy for sequentially allocating treatments to in-trial patients and, because it is hard to compute, propose several computable heuristics based on Bayesian onestep look-ahead techniques. We prove that several of our proposed heuristics are asymptotically optimal in learning treatment strategies.

Numerical results based on two case studies motivated by sepsis management show that our heuristics can significantly improve clinical trial efficiency to learn a treatment strategy for precision medicine.We provide extensions that allow for rewards from outcomes of in-trial patients (resolving the exploration-exploitation trade-off) and for inferring covariate structure using Lasso when biomedical insights on covariate structure are lacking.

Our proposed trial design is of interest to funders, designers, and managers of clinical trials. It may also apply to other contextual bandit problems in settings where insights about covariate-treatment interactions are available.

About the speaker

Spyros Zoumpoulis is a tenured associate professor of Decision Sciences at INSEAD. His research is on developing prescriptive analytics and algorithms for personalization and experimentation, with applications in marketing, healthcare, public health policy, and revenue management. In his work, he develops frameworks and models that quantify the value of experiments for learning and making better personalized decisions, and develops novel algorithms for solving real-world problems.

His research has appeared in leading management science academic journals such as Management Science and Operations Research. He is currently an Associate Editor for Operations Research and for the INFORMS Journal on Data Science.

In his work, Spyros partners with retailers, healthcare providers, and tech companies. He has worked with companies including Microsoft, LinkedIn, IBM, Oracle, and Accenture and has served on the advisory board of start-ups in the areas of his expertise.

At INSEAD, he has taught executive education modules on AI, directs the AI for Boards executive education program, and has taught the MBA electives on foundations of AI for managers and decision models, the MBA and Executive MBA core course on uncertainty, data and judgment, the MBA business foundations course on quantitative methods, the PhD courses on probability and statistics, and the INSEAD-Sorbonne business foundations course on uncertainty, data and judgment. He has won the Dean’s Commendation for Excellence in MBA Teaching award numerous times and has been nominated for the best MBA elective professor award.

Spyros received the B.S., M.Eng., and Ph.D. degrees in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology.

Language

English

This is a free seminar. Registration is mandatory.