Transparent Diabetes Risk Prediction through Interpretable Machine Learning and XAI Integration

Diabetes is one of chronic diseases be afflicted to youth all around the world. This study focuses at the use of various machine learning (ML) methods that works better in type-2 diabetes risk prediction. We developed and evaluated various models, including Random forest, Decision Tree, Logistic Regression, XGBoost, Bagging with different architectures. The study throw light to importance of model’s transparency and Interpretability with Explainable Artificial Intelligence to gain trust in model. We used PIMA dataset with different sampling and cross validations techniques to propose best model with 86% accuracy and 93% AUC-ROC value. The literature review covers recent advancements in Diabetes predictions, classification, care and managements. The results show that proposed model can significantly enhance Diabetes risk prediction and classification with additional trust and explainabilty to improve transparency in model as compared to traditional black box models.