Research in Progress | Rising Star Seminar

Accelerating Drug Discovery and Development with Machine Learning: From Molecules to Patients

Developing a successful therapeutic requires overcoming distinct scientific and computational challenges across multiple stages, from molecular design and preclinical evaluation to clinical studies. Although machine learning has demonstrated remarkable success in individual applications, each stage presents unique data characteristics, modeling assumptions, and translational barriers. In this talk, I will present a series of machine learning approaches developed for different stages of the drug development pipeline. I will begin with monoclonal antibody design, where we developed generative models for de novo antibody design, systematically benchmarked antibody optimization methods, and designed new algorithms for affinity maturation. I will also present a case study on developing broadly neutralizing antibodies against influenza. I will then discuss machine learning approaches for preclinical drug test, focusing on few-shot prediction of synergistic drug combinations using large language models to improve prediction under limited experimental data. Finally, I will present our work on disease subtyping and patient stratification using multimodal clinical data, illustrating how machine learning can improve our understanding of disease heterogeneity and support more precise clinical decision-making. I will conclude by discussing the common opportunities and challenges that emerge across the drug development process to accelerate the development of next-generation therapeutics.

Yejin Kim is a tenured Associate Professor and Associate Director of the Center for Secure Artificial Intelligence for Healthcare in the Department of Health Data Science and Artificial Intelligence. She received her Ph.D. in applied machine learning from Pohang University of Science and Technology (POSTECH), South Korea. Her research focuses on developing artificial intelligence algorithms for therapy development and biomedical discovery. Her work has advanced the field of biomedical AI through innovative machine learning methods and a strong record of externally funded research. Over the past five years, as principal investigator, she has secured more than $8 million in extramural research funding, including two NIH R01 grants. She has published more than 60 peer-reviewed papers in leading biomedical informatics journals, including npj Digital Medicine, JAMIA, as well as premier artificial intelligence and machine learning conferences such as SIGKDD and ACL. Beyond her research, she has made substantial contributions to the scientific community through leadership and professional service. She has served on multiple NIH grant review, including CSR CDMA and HSS, and has reviewed research proposals for funding agencies in the United Kingdom, Canada, and Israel. She serves as an Academic Editor for PLoS Biology, reviews manuscripts for leading journals including Nature Communications and npj Digital Medicine, and has served on the program committees of major AI conferences, including ICLR, NeurIPS, ACL, and AAAI.