Machine Learning-Guided Treatment Discovery

This lecture by the instructor focuses on personalized medicine and the omics revolution, where patients receive targeted therapy tailored to their molecular profile. The talk delves into the use of machine learning to predict treatment outcomes and the challenges faced by traditional algorithms. The instructor presents their research on developing robust machine learning models that encode the nature of the problem as a crucial inductive bias. Additionally, the lecture covers topics such as neural optimal transport, conditional neural optimal transport, and evaluating the effectiveness of these models in predicting treatment responses. Future work includes spatiotemporal modeling of tissue structure and interaction to quantify therapy success.