phenotypic-grouping

This paper introduces REVEAL++, a differentiable phenotypic grouping method for vision-language contrastive learning, applied to retinal fundus images and clinical risk narratives for Alzheimer's disease risk prediction, outperforming discrete grouping baselines.