Enhancing Diabetic Retinopathy Diagnosis with Machine Learning: A Random Forest Approach Using Muscle Response Data

Diabetic retinopathy (DR) is one of the leading causes of loss of vision and blindness worldwide, affecting millions of people each year. As the prevalence of diabetes continues to rise globally, early detection of DR becomes increasingly crucial in preventing irreversible damage to the eyes. This study presents a Random Forest-based machine learning (ML) approach for improving testing and the diagnosis of DR patients using muscle response data. The proposed methodology not only focuses on categorizing patients based on their condition but also extends to classifying the type of visual stimuli administered during tests. This dual classification offers valuable insights into both the patient's physiological condition and the impact of different visual stimuli on muscle responses. The dataset includes measurements from three patient groups: healthy individuals, retinopathic patients, and laser-treated patients, each exposed to five different visual stimuli. Results show that the model effectively classifies both patient categories and stimulus types, revealing meaningful patterns in the relationship between muscle responses and the stimuli. Furthermore, experiments with varying sample sizes underscore the influence of acquisition protocols and sample quantity on classification accuracy. By demonstrating the potential of ML to enhance DR diagnosis, this study opens new possibilities for incorporating ML techniques into clinical settings, offering more personalized diagnostic strategies and improving patient outcomes through more accurate and quick assessments.