Prediction models are abundant in the medical literate, yet very few are implemented and demonstrate clinical impact. This seminar will start with a brief introduction to clinical prediction models before discussing pitfalls and opportunities of applying machine learning methods in this domain. The seminar targets a non-specialist audience and introduces concepts via examples using different data modalities (images, voice, etc.) from cardiometabolic research. Finally, the seminar will provide perspectives on how AI is perceived by patients and the general public.

Bio: Adam Hulman is a senior researcher leading the Machine Learning & Clinical Prediction Lab at Steno Diabetes Center Aarhus, Aarhus University Hospital, and an associate professor at the Department of Public Health, Aarhus University. Adam is an applied mathematician by training with a PhD in diabetes epidemiology. His lab’s overarching goal is to turn health data into clinical insights and applications by using advanced statistical and machine learning methods. More specifically, the group works on the development and application of deep learning methods to be able to integrate clinical data of different types (tabular, images, time series, voice) in risk prediction of diabetic complications. He is committed to hearing the users’ voices about AI and to building bridges between diabetes researchers, clinicians, and the data science community.