Visual Acuity Classification Using Ensemble Machine Learning Algorithms: With Explainable Approach

Visual acuity is an important aspect of overall health, and early detection of visual problems can prevent long-term complications. This study explores the use of ensemble machine learning algorithms to predict visual acuity categories before they happen. We compare the performance of different types of ensemble algorithms, AdaBoost, XGBoost, Gradient Boosting, Stacking and Bagging Classification accuracy is 0.99, 0.98, 0.99, 0.98 and 0.99 respectively. Parameter optimization is also used to find the best set of hyper-parameters for a model that will help it perform optimally. In addition, we investigate the impact of various attributes, including living conditions, age, gender, and source of drinking water, on visual acuity classification. We employed SHAP for model interpretation and explanation. Our results revealed that these factors can indeed be risk factors for blindness or visual impairment, and that incorporating them into the prediction model can improve accuracy and early detection. Our research offers for predicting and preventing visual acuity categories. By identifying individuals at risk, we can implement timely interventions and treatments, consequently improving the quality of life for those affected.