Deep Kernel Learning for Stratifying Glaucoma Trajectories

arXiv cs.LG / 5/4/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper tackles the clinical challenge of stratifying glaucoma progression risk using sparse, irregularly sampled multimodal EHR data.
It proposes a deep kernel learning (DKL) model that combines a transformer-based feature extractor using clinical-BERT embeddings with a Gaussian Process (GP) backend.
The approach identifies three clinically distinct patient subgroups and demonstrates that it can separate progression risk from current disease severity.
It finds a high-risk group with a worsening trajectory even though their average visual acuity is better than that of another stable-poor subgroup, indicating progression-focused stratification.
The authors position the method as a decision-support tool that could enable targeted interventions for high-risk glaucoma patients and improve care management.

Abstract

Effectively stratifying patient risk in chronic diseases like glaucoma is a major clinical challenge. Clinicians need tools to identify patients at high risk of progression from sparse and irregularly-sampled electronic health records (EHRs). We propose a novel deep kernel learning (DKL) architecture that leverages a Gaussian Process (GP) backend. The GP's kernel is defined by a transformer-based feature extractor applied to clinical-BERT embeddings to model glaucoma patient trajectories from multimodal EHR data. Our method successfully identifies three clinically distinct patient subgroups. Crucially, the model learns to decouple disease progression from current severity, identifying a high-risk group with a worsening trajectory despite having better average visual acuity than a second, stably poor group. This reveals that the model learns to identify progression risk rather than just the current disease state. This ability to stratify patients based on their risk trajectory progression offers a powerful tool for clinical decision support, enabling targeted interventions for high-risk individuals and improving the management of glaucoma care.