Fostering trust and interpretability: integrating explainable AI (XAI) with machine learning for enhanced disease...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Machine Learning for Disease Prediction

Background:

• Artificial Intelligence (AI) significantly advances early disease detection and clinical decision support.
• The 'black box' nature of AI models hinders widespread adoption in healthcare due to a lack of transparency.
• Medical practitioners require understandable reasoning for AI-driven diagnostic outcomes.

Purpose of the Study:

• To develop a hybrid Machine Learning (ML) framework integrating Explainable AI (XAI) strategies.
• To enhance both the predictive performance and interpretability of AI models in healthcare.
• To address the critical need for transparency in AI-driven medical diagnostics.

Main Methods:

• Utilized a hybrid ML framework incorporating Decision Trees, Naive Bayes, Random Forests, and XGBoost algorithms.
• Integrated Explainable AI (XAI) techniques, specifically SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations).
• Applied the framework for predicting risks of Diabetes, Anaemia, Thalassemia, Heart Disease, and Thrombocytopenia.

Main Results:

• Achieved a high accuracy of 99.2% in predicting various medical conditions.
• The framework successfully provided understandable explanations for AI-generated predictions.
• Key features contributing to each prediction were identified and displayed using SHAP and LIME.

Conclusions:

• The developed framework enhances AI model interpretability and predictive accuracy in healthcare.
• Understandable AI outputs empower clinical practitioners to make informed decisions, reducing distrust.
• This approach bridges the gap between AI capabilities and clinical adoption in critical medical scenarios.