Related Experiment Videos
An Interpretable Fuzzy Distance-Based Ensemble Framework with SHAP Analysis for Clinically Transparent Prediction of
Asif Hassan Syed1, Altyeb Altaher Taha2, Ahmed Hamza Osman3
1Department of Computer Science, Faculty of Computing and Information Technology, Rabigh, King Abdulaziz University, Jeddah 21589, Saudi Arabia.
Diagnostics (Basel, Switzerland)
|May 13, 2026
Summary
This study introduces an interpretable AI model for early diabetes prediction, achieving high accuracy and providing clinicians with transparent, actionable insights for better patient care.
Area of Science:
- Artificial Intelligence in Medicine
- Machine Learning for Healthcare
- Clinical Decision Support Systems
Background:
- Diabetes mellitus is a global chronic metabolic health concern.
- Early prediction of diabetes is crucial for mitigating disease severity.
- Existing prediction models often lack transparency and clinical interpretability.
Purpose of the Study:
- To develop an interpretable ensemble model for accurate diabetes prediction.
- To enhance clinical trust through transparent AI decision-making.
- To provide calibrated confidence scores for risk stratification.
Main Methods:
- An interpretable multi-metric fuzzy distance-based ensemble (MMFDE) was developed.
- MMFDE integrates gradient-boosting classifiers using a novel fuzzy fusion mechanism.
- Predictions are transformed into a fuzzy space using hybrid distance metrics and SHAP analysis for feature importance.
Main Results:
- MMFDE achieved 94.83% accuracy and 97.66% AUC on the HFGDD dataset.
- The model demonstrated high performance and interpretability, outperforming base classifiers.
- Identified key risk factors: glucose, BMI, and diabetes pedigree function.
Conclusions:
- MMFDE offers a trustworthy AI solution for clinical decision support in diabetes prediction.
- High performance and interpretability are achievable simultaneously in AI healthcare models.
- The system supports automated screening and clinician review workflows.
