Georgia Perakis

Massachusetts Institute of Technology

Professor, Operations Management and Operations Research and Statistics

PLCO (Learn more about this study)

PLCOI-1256

Jun 22, 2023

Impact of treatment options on patient outcomes in hard - to - treat cancers: creating a personalized treatment framework using Machine Learning to address disparities in developed v. LMIC nations’ healthcare

We propose to develop a multi-modal framework to assist in decision support of the optimal treatment strategy for individual patients. Our framework consists of a deep learning predictive model (from the images) and an optimization model. The predictive model takes as input patient-specific information, clinical data, imaging data, and treatment costs in different countries (developed world versus low-and middle-income (LMIC) countries) and predicts theoutcome for a given treatment plan.The target function being used is the tumor response criteria as defined by RECIST (response evaluation criteria in solid tumors) 1.1 guidelines (Eisenhauer et al., 2009). Building on the framework developed by Kumar and co-workers(Chen et al., 2021), we aim to optimize the treatment plan based on clinical targets and predicted treatment outcome.