Jian Li, an assistant professor at Stony Brook University, has received a $1.2 million R01 grant from the National Institutes of Health to develop artificial intelligence systems to improve the allocation of healthcare resources for patients with diabetes, cardiovascular disease, and maternal health concerns. The four-year grant is part of the NSF/NIH Smart and Connected Health program, supporting research at the intersection of technology, medicine, and public health.
Li, also a core faculty member of Stony Brook’s AI Innovation Institute, is leading the project as a multiple principal investigator with Kai Wang from Georgia Institute of Technology. Researchers from Harvard Medical School and Massachusetts General Hospital are also collaborating, combining expertise in AI, clinical practice, and public health.
“Healthcare workers face an enormous challenge: how do you allocate limited resources to help the most people?” Li said. “Our goal is to develop intelligent systems that can learn, adapt, and make fairer decisions about who receives interventions and when.”
The research addresses complex public health challenges, especially where budgets are limited and resource distribution decisions are critical. The team is using advanced machine learning methods called Restless Multi-Armed Bandits to make sequential decisions under uncertainty while considering patient diversity and social factors. These algorithms help healthcare systems prioritize interventions in ways that are effective and equitable.
To support the work, researchers will use health datasets, including electronic health records from MIMIC-III, MIMIC-IV, the NIH All of Us Research Program, and the Mass General Brigham Biobank. Data from intervention trials such as REAL HEALTH-Diabetes, Look AHEAD, and the Diabetes Prevention Program will ensure the AI systems apply to real-world clinical scenarios. By integrating advanced AI with rich healthcare data, the project aims to create tools that enhance decision-making, reduce disparities, and improve patient outcomes across diverse populations.