Title: Knowledge-informed Machine Learning in health applications
Jing Li, Georgia Institute of Technology
Dr. Jing Li is the Virginia C. and Joseph C. Mello Chair and Professor in the H. Milton Stewart School of Industrial and Systems Engineering, and a core faculty of the Machine Learning Center at Georgia Tech. Her research interests are data fusion and machine learning intersecting with domains having complex data structures in health/medicine, manufacturing, and agriculture. Her research received various Best Paper awards from IISE, INFORMS, and medical societies. Prior to joining Georgia Tech, she was on the faculty of Arizona State University and a co-founder of the ASU-Mayo Clinic Center for Innovative Imaging. She is currently a Senior Editor for IEEE Transactions on Automation Science and Engineering and a Department Editor for IISE Transactions on Healthcare Systems Engineering. She receives funding support from NIH, NSF, DOD and industry, and is an NSF CAREER awardee. She is a fellow of IISE.
Abstract
In the rapidly evolving era of AI, does human domain knowledge still matter in driving discoveries? This talk focuses on knowledge-informed machine learning (ML) in health applications. I will begin by highlighting some key findings in our recent review paper on this topic. Then, I will present a collection of our research works that integrate domain knowledge of various forms, such as simulation models, weak labels, hierarchical relationships, qualitative descriptions, and foundation models and LLM prompting, into ML/AI model design. Our research applications span cancer, Alzheimer’s disease, neurological disorders, and dental health. I will conclude the talk by pointing out some future directions.