A Survey on Medical Explainable AI (XAI): Recent Progress, Explainability Approach, Human Interaction and Scoring...

Area of Science:

• Medical Artificial Intelligence
• Explainable AI (XAI)

Background:

• The integration of Artificial Intelligence (AI) in healthcare necessitates explainability for legal and ethical compliance.
• Understanding patient conditions and treatment decisions requires transparent AI models.

Purpose of the Study:

• To present a comprehensive survey of eXplainable AI (XAI) in the medical domain.
• To explore model enhancements, evaluation methods, case studies, and datasets relevant to medical XAI.
• To propose advancements in XAI feedback systems and scoring mechanisms for healthcare applications.

Main Methods:

• Review of existing AI and XAI methodologies, including local/global methods, knowledge distillation, and interpretable machine learning.
• Analysis of practical case studies to demonstrate current XAI progress in medicine.
• Development of a user-in-the-loop approach emphasizing human-machine collaboration.

Main Results:

• Identification of key XAI characteristics and future trends in healthcare explainability.
• Insights into prerequisite considerations for initiating medical XAI projects.
• Introduction of a novel XAI recommendation and scoring system to address limitations in current evaluation methods.

Conclusions:

• The survey highlights the critical need for explainable solutions in high-impact medical applications.
• A human-in-the-loop approach with enhanced feedback mechanisms can improve the reliability of medical XAI.
• Further development and implementation of XAI are essential for advancing ethical and effective AI in healthcare.