Biostatistics and Medical Informatics Seminar with Lynn Lin, PhD

Modern AI in biomedicine depends critically on learning useful representations. In this talk, Dr. Lin will present two complementary statistical learning strategies for moving from cells to patients beyond coarse pseudo-bulk summaries. The first learns predictive patient representations from individual cells, illustrated by cytoGPNet. The second represents each patien tas a distribution over cells, enabling interpretable and principled comparisons of whole cell populations.