AN EXPLAINABLE ARTIFICIAL INTELLIGENCE FRAMEWORK FOR EARLY PREDICTION OF DIABETES MELLITUS USING CLINICAL DATA

Abstract

Diabetic mellitus that induces a high-burden disease, emerging morbidity. Risk stratification enables early intervention to mitigate disease progression and rapid therapeutic intervention. Machine learning frameworks have exhibited enhanced discriminative capability; proven methodologies utilize black-box model, algorithmic reductionism. This study demonstrates an explainable, automated diabetes prediction system leveraging clinical data. The proposed framework integrates data preparation, data transformation, and meta-algorithms. To verify auditability and interpretability, explainable AI models such as agnostic explainers are converging to enable context-aware analytics. This study highlights clinical judgement, promoting health literacy and explainability. The architecture is empirically validated by performance metrics like accuracy, precision, repeatability, unbiased accuracy, and separation effectiveness. Data-driven proof establishes that the proposed yields produce superior predictive performance against interpretable baselines. The data suggest that interpretable machine learning with data mining enhances both validity and workflow integration. This study facilitates the enhancement of reliable and explainable AI for diabetic risk assessment.