Current methods in explainable artificial intelligence and future prospects for integrative physiology

Area of Science:

• Physiological research
• Medical informatics
• Artificial intelligence

Background:

• Artificial intelligence (AI) is increasingly used in physiological research for analysis and prediction.
• Explainable artificial intelligence (XAI) aims to make AI models transparent and understandable to human decision-makers.
• XAI methods include inherently interpretable models and post hoc explanations for opaque models.

Purpose of the Study:

• To introduce core topics and current XAI methods relevant to physiology.
• To review solved and open challenges in XAI research within the medical field.
• To provide an outlook on future applications of XAI in integrative physiological research.

Main Methods:

• Selective overview of current XAI methods in physiology.
• Illustration of practical examples from the medical field.
• Discussion of challenges and future prospects in XAI research.

Main Results:

• XAI is crucial for understanding AI in physiological research.
• Current XAI methods offer ways to interpret complex AI models.
• Practical examples highlight both successes and challenges in applying XAI.

Conclusions:

• XAI is essential for developing trustworthy AI in integrative physiological research.
• Future directions involve integrating physiological expertise into XAI development for better human-AI partnerships.
• XAI promises to enhance the utility and benefit of AI in medical and physiological applications.