Role of Explainable AI in Medical Diagnostics and Healthcare: A Pilot Study on Parkinson's Speech Detection

Decision making is one of the most important steps in healthcare applications. There are lot of Artificial Intelligence (AI) based tools have been developed to provide better solutions for diagnostic problems in last two or three decades. In hospitals environment, the trust on the results provided by these AI models is a question mark as they are simply a black box to healthcare professionals. This is due to the lack of explainability and interpretability of the results provided by those models. Explainable AI is a rapidly growing technology to address this issue by incorporating various methods to bring explainability and interpretability. This paper aims at providing a comprehensive review on explainable AI methods addressed for health care applications. Furthermore, a pilot study on the use of Explainable AI algorithms using Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP) algorithms is experimented on Parkinson's disease detection.