Deep Kernel Learning for Stratifying Glaucoma Trajectories
Bruce Rushing, Angela Danquah, Alireza Namazi, Arjun Dirghangi, Heman Shakeri · 2026
Read Deep Kernel Learning for Stratifying Glaucoma Trajectories by Bruce Rushing, Angela Danquah, Alireza Namazi, Arjun Dirghangi, Heman Shakeri (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-00708.txt.
Description
Bruce Rushing and 4 co-authors stratify glaucoma patients by where their disease is heading, not just how bad it is today. Clinicians managing this chronic eye disease need to spot who will progress, from sparse, irregularly sampled health records. Their architecture pairs deep kernel learning with a Gaussian process back end, whose kernel is built by a transformer over embeddings of the clinical record produced by a medical language model.
The model sorts patients into three clinically distinct subgroups. Its sharpest finding is a decoupling: one high-risk group is worsening steadily while still seeing better, on average, than a second group whose vision is stably poor. A severity-only tool would rank those groups backwards. By learning trajectory risk rather than current state, the system can flag quietly deteriorating patients for targeted intervention. The authors offer it as decision support for glaucoma care.
Catalog record
| CREATOR | Bruce Rushing, Angela Danquah, Alireza Namazi, Arjun Dirghangi, Heman Shakeri |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50 KB |
| LICENSE | http://creativecommons.org/licenses/by-sa/4.0/ |
| RIGHTS | CC BY SA |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.00708