REVEAL++: Differentiable Phenotypic Grouping for Vision-Language Retinal Modeling of Alzheimer's Disease Risk

2026-06-17

AI Summary

This work addresses the limitations of existing retinal–language joint modeling approaches for Alzheimer’s disease risk prediction, which rely on discrete phenotypic groupings that yield rigid supervision signals decoupled from representation learning. To overcome this, the authors propose a differentiable framework that models continuous phenotypic similarity through soft-weighted multi-positive contrastive learning, transforming hard group assignments into learnable continuous signals for end-to-end optimization of cross-modal alignment and phenotypic structure. The method leverages intra-modal embedding similarities between retinal images and clinical risk profiles to construct differentiable weighting functions and continuous aggregation operators. Experiments on the UK Biobank demonstrate that the proposed approach significantly outperforms conventional discrete-group contrastive learning and standard vision–language baselines, achieving improved performance in Alzheimer’s disease risk prediction.

Abstract

The retina offers a noninvasive window into neurodegenerative disease, capturing subtle structural patterns associated with a risk of future cognitive decline. Vision-language alignment frameworks such as REVEAL have shown that pairing retinal fundus images with structured clinical risk narratives improves early prediction of Alzheimer's disease (AD). A key design choice in these approaches is the use of phenotypic grouping, where individuals with similar risk profiles are treated as multi-positive pairs during contrastive learning. However, existing methods operationalize phenotypic similarity as a discrete construct, relying on hard group assignments that impose rigid supervision and decouple group formation from representation learning. We propose a continuous formulation of phenotypic structure within contrastive learning. Rather than assigning samples to fixed clusters, we model inter-subject similarity as a differentiable weighting function derived from intra-modality embedding similarities in both retinal images and risk profiles. These weights define soft multi-positive relationships through a continuous aggregation operator, enabling graded supervision that reflects the spectrum nature of disease risk. We further introduce a soft-target contrastive objective that jointly learns cross-modal alignment and phenotypic structure in an end-to-end manner. Evaluated on UK Biobank retinal imaging data for incident AD prediction, the proposed framework consistently outperforms discrete group-based contrastive learning and standard vision-language baselines. By treating phenotypic similarity as a learnable, continuous signal rather than a fixed grouping rule, our approach provides a principled and robust foundation for population-scale neurodegenerative risk modeling from multi-modal retinal and clinical data.

Problem

• phenotypic grouping
• Alzheimer's disease
• contrastive learning
• retinal imaging
• vision-language modeling

Innovation

• differentiable phenotypic grouping
• vision-language modeling
• contrastive learning
• retinal imaging
• Alzheimer's disease risk prediction