LMT: Longitudinal Mixing Training for Disease Progression Prediction

Abstract

LMT enables predicting disease progression from a single image at inference time by leveraging longitudinal data mixing during training across multiple sources.