| Abstract |
ML/DL method are widely used for prediction in biomedical research, but predictive performance alone often provides limited scientific insight. Explainable artificial intelligence (XAI) and visualization methods can help identify important variables, characterize complex relationships, and generate interpretable findings beyond prediction accuracy.
In this seminar, I will introduce practical XAI and visualization approaches and demonstrate their applications using two biomedical examples: drug-related expression prediction in human and mouse liver data, and analysis of hypnotic medication timing and use patterns. These examples illustrate how predictive modeling can be extended toward interpretation, hypothesis generation, and data-driven scientific discovery. |
