|
Bibhas ChakrabortyAssistant Professor
722 West 168th Street, 6th Floor |
Prior to joining Columbia, I completed my PhD from the Department of Statistics at the University of Michigan, Ann Arbor, under the supervision of Prof. Susan A. Murphy. Here is a copy of my dissertation.
Dynamic Treatment Regimes: My primary research interest falls in the area of dynamic treatment regimes or adaptive treatment strategies. Dynamic treatment regimes are decision rules about recommended treatments based on past treatment and time-varying patient characteristics. Once developed, they can be employed to enhance the clinical judgments used in practice. I work on developing methods of estimation for these decision rules and valid inference about them from longitudinal data. In addition, I am interested in designing sequential multiple assignment randomized trials (SMARTs) that give rise to high-quality data that can be used to develop and optimize dynamic treatment regimes.
Reinforcement Learning & Machine Learning: A related area of interest is reinforcement learning, a branch of machine learning where an agent learns to choose optimal actions by interacting with the environment. This has striking similarity with the problem of estimating optimal dynamic treatment regimes. I work on modification of classical reinforcement learning algorithms for medical applications. I am also interested in machine learning and data mining in general, and have recently collaborated with researchers at the Columbia Center for Computational Learning Systems.
Design of Clinical & Behavioral Intervention Trials: My another research thread lies in design of experiments for developing multicomponent interventions, often involving behavioral or delivery components (factors). In particular, I have worked on the use of fractional factorial designs for such trials. I was involved in the design and analysis of a smoking cessation trial for cancer prevention involving multicomponent behavioral interventions. I am also interested in adaptive clinical trials and have done some work in this area.