Machine Learning Flags Key Risks for Early Macular Degeneration: Hypertension, Joint Disorders, and Dyslipidemia: Study
USA: A new study published in Investigative Ophthalmology & Visual Science and presented at the 2025 ARVO Annual Meeting has identified critical comorbidities that may help predict early-onset age-related macular degeneration (AMD). Conducted by Ethan Wu and colleagues from the University of Pittsburgh School of Medicine, the study leveraged machine learning to examine how systemic health conditions influence the early development of AMD—a major cause of visual loss among the aging population.
The study analyzed data from 930 individuals diagnosed with AMD, categorizing them into two groups: early-onset (before age 65; n=392) and late-onset (after age 85; n=538). To enable standardized comparisons, only comorbidities identified before age 55 were included. The researchers extracted demographic and health information from electronic medical records to construct detailed patient profiles.
Using unsupervised clustering via Uniform Manifold Approximation and Projection (UMAP), the team identified three distinct patient clusters based on comorbidity patterns. Cluster 1 had the highest proportion of early-onset AMD cases (76.09%) and was notably marked by inflammatory joint disorders (60.1%) and hypertension (34.1%). Cluster 2, which also had a high burden of early-onset AMD (69.83%), showed a significant association with hypertension (25.9%). In contrast, Cluster 3 had the lowest rate of early-onset cases (27.02%) and presented minimal or no comorbidities.
To assess predictive potential, the researchers applied machine learning models—including gradient-boosted decision trees and random forests—to comorbidity data. The gradient-boosted decision tree model achieved a prediction accuracy of 75.84%, relying solely on comorbidities to determine the likelihood of early AMD onset.
In analyzing which comorbidities were most predictive, hypertension emerged as the top-ranked factor, followed by dyslipidemia and inflammatory joint disorders. These findings highlight the potential of systemic health factors in influencing the earlier manifestation of AMD.
The study also found that patients with early-onset AMD had a significantly higher average number of comorbid conditions compared to those with late-onset AMD (3.32 ± 0.24 versus 0.16 ± 0.02), indicating a potential role of overall disease burden in accelerating retinal degeneration.
The authors concluded that hypertension, dyslipidemia, and inflammatory joint disorders are strongly associated with early AMD onset. They emphasized the need for ophthalmologists to consider patients’ systemic health profiles during routine evaluations. Integrating comorbidity screening and proactive management may enable earlier diagnosis and personalized interventions to slow disease progression.
This innovative study emphasizes the promise of machine learning in ophthalmology and highlights the importance of interdisciplinary approaches in tackling complex, age-related diseases like AMD.
Reference:
Ethan Wu, Nasiq Hasan, Sharat Chandra Vupparaboina, Jessica Ye Jiang, Joseph DeCicco, Kiran Vupparaboina, Sandeep Bollepalli, José-Alain Sahel, Jay Chhablani; Identifying Key Co-Morbidities for Predicting Early-Onset Age-Related Macular Degeneration Using Machine Learning. Invest. Ophthalmol. Vis. Sci. 2025;66(8):293.
