Human-centered AI Approaches for Individualized Self-management Regimens
Date: Tuesday, October 11, 2022
Time: 3:00 p.m. to 4:00 p.m. ET
Type of event: Ada Lovelace Computational Health Lecture
Location: Virtual; View through NIH Videocast
Noémie Elhadad, PhD is an Associate Professor of Biomedical Informatics, affiliated with Computer Science and the Data Science Institute at Columbia University. She serves as Vice Chair for Research and Graduate Program Director for the Department of Biomedical Informatics (DBMI). She leads Even, the Data-Powered Women's Health Research Initiative at Columbia University as well as the Citizen Endo project, which advances research in endometriosis through citizen science. Dr. Elhadad’s research interests are at the intersection of machine learning, natural language processing, medicine, and technology. She investigates ways in which observational clinical data (e.g., electronic health records) and patient-generated data (e.g., online health community discussions, mobile health data) can enhance access to relevant information for clinicians, patients, and health researchers alike and can ultimately impact healthcare and health of patients.
Prior to joining Columbia DBMI in 2007, Dr. Elhadad completed her PhD in Computer Science at Columbia University and was an Assistant Professor in Computer Science at The City University of New York.
Personal health informatics solutions have been proposed to support self-management, to scaffold problem solving for individuals, and to promote experimentation that help identify potential triggers of disease flares across a range of health conditions. In many chronic diseases however, there is strong evidence of person-to-person variation in treatment responses and associated symptoms. In addition, there are often no predetermined policy guidelines for self-management, and if there are, individuals are left with the burden of translating them into their day-to-day lives. In this talk, I will discuss the challenges and exciting research directions for augmenting personal health informatics systems with AI-driven recommendations for self-management strategies. Because self-management is more successful when aligned with an individual's goals and context of daily living, as well as with their own health status and physiological responses, I argue that the promise of automated recommendations hinges on their personalization and posit that reinforcement learning is a promising technique for learning and delivering such personalized self-management recommendations, if designed in a human-centered fashion.