Explainable Medical Diagnosis System Using Machine Learning and Explainable AI

25 Jul
Posted on Author

Authors: Navya Shree K V, Sagar D, H S Shashank, Sinchana S Y, Professor Dr. Dilshad Begum, Professor Dr. T Venkatesh

Abstract: Artificial Intelligence (AI) has significantly improved healthcare by enabling accurate and timely disease diagnosis; however, the limited interpretability of deep learning models restricts their adoption in clinical practice. This paper proposes an Explainable Medical Diagnosis System (EMDS) that integrates machine learning, deep learning, and Explainable Artificial Intelligence (XAI) to deliver accurate, reliable, and interpretable disease prediction. The proposed framework consists of six modules: medical data acquisition, data preprocessing, feature engineering and selection, hybrid disease classification, explainable AI analysis, and clinical decision support. During preprocessing, missing values are imputed, medical images are enhanced using Contrast Limited Adaptive Histogram Equalization (CLAHE), and class imbalance is addressed using the Synthetic Minority Oversampling Technique (SMOTE). Feature optimization is performed through Recursive Feature Elimination (RFE) and Mutual Information. The hybrid classification model combines EfficientNetV2 for deep image feature extraction, TabNet for structured electronic health record learning, and XGBoost for disease classification, while a stacking ensemble improves predictive accuracy and generalization. Model interpretability is achieved using SHapley Additive exPlanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), and Grad-CAM++, providing both global feature importance and patient-specific visual explanations. Experimental evaluation using benchmark medical datasets and five-fold cross-validation demonstrates superior performance over conventional approaches in terms of accuracy, precision, recall, F1-score, and AUC. The proposed EMDS provides a scalable, interpretable, and trustworthy framework for AI-assisted clinical decision-making in modern healthcare systems.