An Explainable Ai Model for Diabetes Prediction Using Random Forest

Diabetes is a chronic metabolic disorder caused by insufficient insulin production or ineffective insulin utilization, resulting in elevated blood glucose levels. As a major global health concern, it affects millions and can lead to severe complications if undiagnosed or mismanaged. This study proposes an explainable artificial intelligence model for accurate diabetes prediction using a calibrated Random Forest classifier. To ensure reliability, comprehensive data preprocessing techniques were implemented, including regression-based imputation, Recursive Feature Elimination, outlier removal using the IQR method, and class imbalance handling with ADASYN. Hyperparameter tuning through RandomizedSearchCV and model calibration improved predictive accuracy. The final model achieved an accuracy of 88 %, precision of 87 %, recall of 87 %, and F1-score of 86 %. Beyond performance, explainability was prioritized using LIME and decision path visualizations, providing clinicians with transparent reasoning behind predictions. This research demonstrates how integrating XAI techniques with robust machine learning can enhance clinical decision support systems. It also highlights the model's potential for integration into practical web-based tools that assist early detection and improve patient outcomes in diabetes care.