Algorithms for Personalizing Digital Interventions
A presentation from 2022–2023 Radcliffe Alumnae Professor Susan A. Murphy
Susan A. Murphy, the Mallinckrodt Professor of Statistics and of Computer Science at Harvard University, conducts research on how to improve sequential, individualized decision making in health. Her lab develops clinical trial designs and learning algorithms for use in personalizing the delivery of behavioral interventions via smart devices such as smartphones and other wearables. The focus is on helping individuals who are struggling with chronic disorders. She developed the clinical “micro-randomized trial,” which is in use across a broad range of health, including in treatment for substance use addiction, cardiovascular health, take-home chemotherapy, and mental illness.
During her Radcliffe fellowship, Murphy is working on developing and improving algorithms for personalizing digital interventions. These algorithms need to learn quickly when and where it is best to deliver the interventions so that individuals do not become overburdened and discouraged. The algorithms will appropriately pool data from different individuals to learn more quickly, in real time, on each individual. In the language of statistics, this requires careful analysis of the bias-variance tradeoff.