H-COR Works in Progress Seminar - Dr. Vibhuti Gupta

Generalizable and Interpretable Multimodal AI Fusion Framework for Precision Medicine Using Foundation Models 

With the rapid growth of big data in precision medicine, high-dimensional multimodal biomedical data are routinely generated in clinical practice. Integrating these heterogeneous data including medical images, clinical notes, lab tests, genomics, and EHR data using machine learning and deep learning can uncover genotype–phenotype relationships, improve patient stratification, and support timely clinical decision-making. However, the scale and heterogeneity of biomedical datasets across specialties make effective integration challenging. Foundation models provide a promising solution by learning generalizable representations from large, diverse datasets that can support many downstream clinical tasks. In this talk, Dr. Vibhuti highlights prior work on multimodal AI frameworks for prostate cancer risk prediction and discusses current research on leveraging foundation models to develop generalizable and interpretable multimodal approaches for precision medicine.