Comparative Analysis of Deep Learning Models for Automated Cataract Detection in Medical Imaging

Lung disease remains a major global health concern, requiring early and accurate diagnosis for effective treatment. This study presents a novel approach for lung disease classification using Principal Component Analysis (PCA) for feature reduction and XGBoost for classification. SMOTE addresses class imbalance, while Min-Max normalization standardizes data for improved model performance. PCA reduces dimensionality, pre- serving essential features and enhancing computational efficiency. The proposed model achieves 96.5% accuracy, outperforming traditional classifiers in precision and recall. The novelty lies in combining feature extraction, synthetic data generation, and ensemble learning to create a more reliable diagnostic tool. This system can serve as a valuable clinical support tool, aiding clinicians with fast and accurate lung disease identification. Future work includes integrating deep learning models and developing real-time, cloudbased systems for remote diagnostics, making this approach highly scalable and impactful in resource- constrained healthcare environments.