Trusting AI made decisions in healthcare by making them explainable.

Area of Science:

• Digital Health
• Machine Learning
• Artificial Intelligence

Background:

• Trust issues hinder machine learning integration in mHealth.
• Black-box algorithms lack transparency in their reasoning.
• Explainable AI is crucial for advancing digital health solutions.

Purpose of the Study:

• To develop a transparency layer for black-box machine learning algorithms.
• To enhance the efficiency and trustworthiness of mHealth applications.
• To facilitate the integration of AI in healthcare.

Main Methods:

• Utilized a machine learning testing framework for knowledge transfer.
• Implemented knowledge transfer from a white-box model to a black-box model.
• Evaluated the success of knowledge transfer through correlation analysis.

Main Results:

• Demonstrated distinct reasoning differences between base and knowledge-infused white-box models.
• Achieved a very high correlation between the base black-box model and the new knowledge-infused model.
• Validated the success of the knowledge transfer process.

Conclusions:

• Transparency is essential for digital health and healthcare.
• Explainable AI tools can reduce the obscurity of black-box models.
• This approach contributes to building trust in AI-driven health applications.