Predicting Diabetic Retinopathy with Machine Learning: Disentangling the Diabetes-Ocular Health Complex

This research paper investigates the intricate connections between diabetes and its impact on various organ systems, with a specific focus on ocular health. Leveraging a comprehensive dataset and employing machine learning models, the study delves into factors influencing the need for spectacles in individual’s post-diabetes detection. Notably, age emerges as a significant factor, with those aged 25-70 exhibiting a higher incidence of requiring corrective lenses. Furthermore, the prevalence of Type II diabetes is discernibly correlated with individuals aged 25-54, emphasizing the complex interplay between diabetes subtype and age. Elevated blood glucose levels, specifically Sugar_B levels, demonstrate a positive association with the likelihood of needing spectacles. The research uncovers trends related to blood pressure levels, irritation, cholesterol, light intensity, and strain, shedding light on their implications for visual health post-diabetes detection. The study’s modeling approach, including Logistic Regression and SVM, achieves an impressive $\mathbf{9 2 \%}$ accuracy in predicting diabetic retinopathy. The paper underscores the importance of a holistic approach to diabetes management and eye health, offering insights that can inform clinical practice, public health initiatives, and ongoing research endeavors.