Summary:
UCLA researchers in the Department of Computational Medicine have developed a novel machine learning-based clinical support tool to personalize first-line glaucoma treatment.
Background:
Glaucoma is a heterogeneous, progressive eye disease that is a leading cause of irreversible blindness. Despite the emergence of improved diagnostic tools, treatment decisions for glaucoma remain largely empirical and non-standardized. Clinicians often initiate treatment with medication and may later escalate to combination therapy, laser, or surgery if first-line treatments prove unresponsive. While these methods aim to lower the primary modifiable risk factor for glaucoma progression, there are no validated methods for categorizing patients into subgroups that tailor treatment based on individual characteristics. As a result, patients often endure cycles of medication changes, ineffective therapy, or invasive surgical intervention. This trial-and-error approach contributes to added costs, poor prognosis, and increased burden to all parties involved. There remains an unmet need for a personalized approach that can stratify patients and enable individualized treatment pathways, ultimately improving outcomes, reducing complication risk, and lowering the overall burden on the healthcare system.
Innovation:
To address these limitations, researchers at UCLA have developed a patient stratification machine learning algorithm designed to predict treatment response in patients with open-angle glaucoma. The algorithm utilizes routinely available clinical data to create a patient stratification score at the time of diagnosis, enabling clinicians to rapidly match patients with an optimal first-line treatment. This approach shifts glaucoma care from an empirical, trial-and-error methodology to a personalized regimen, reducing the likelihood for costly medication changes and ineffective cycles of therapy. Early implementation of optimal glaucoma treatment approaches significantly improves long-term medication adherence and reduces the likelihood of disease progression. Additionally, the technology may be directly integrated into existing clinical decision systems, deployable as a customizable, standalone web application. By transforming standard clinical data into actionable treatment insights, this innovation provides a scalable solution that modernizes glaucoma management.
Potential Applications:
● Point-of-Care cinical decision support
● Electronic health record integration
● Clinical trial enrichment
● Telehealth and remote monitoring
● Platform technology for a diverse array of ophthalmologic applications
Advantages:
● Data-driven methodology
● Cost reduction
● Improved long-term outcomes
● Interpretable
● Flexible deployment
Development-To-Date:
First successful demonstration of the invention completed.
Related Papers:
● Rahmani, E., et al. (2025). Epigenetic patient stratification reveals a sub-endotype of type 2 asthma with altered B-cell response. medRxiv, 10.1101/2025.08.28.25334696v1.
● Rahmani, E., et al. (2024). Accurate prediction of disease-risk factors from volumetric medical scans by a deep vision model pre-trained with 2D scans. Nature Biomedical Engineering (Indexed in PubMed: 39354052)
Reference:
UCLA Case No. 2026-122
Lead Inventors:
Elior Rahmani, Arush Ramteke