Performance Evaluation of Explainable Machine Learning on Non-Communicable Diseases

Abstract

The advancements in machine learning and artificial intelligence can significantly benefit the diagnosis of Non-Communicable Diseases (NCDs). However, the inherent complexity of black-box models hinders the interpretability of the model. Potential regulatory issues arise, and the lack of trust within the medical community is apparent due to the lack of understanding of how and why a model made a prediction. In this study, we demonstrate how model-agnostic methods of eXplainable AI (XAI) can help provide explanations to understand black-box models on NCDs datasets better.