Linking Explainability, Trust, and Use: A Framework for Clinical Decision Support

Area of Science:

• Health Informatics
• Artificial Intelligence in Medicine
• Human-Computer Interaction

Background:

• Limited adoption of AI-based clinical decision support systems (AI-CDSS) in healthcare settings.
• Insufficient clinician trust in AI models, particularly in high-risk medical scenarios, hinders AI-CDSS implementation.
• Need for a framework to enhance trust and promote the use of AI-CDSS.

Purpose of the Study:

• To develop an initial conceptual model that links AI explainability and reliability to clinician trust.
• To investigate the relationship between clinician trust and the intention to use AI-CDSS.
• To provide guidance for designing trustworthy AI-CDSS for healthcare.

Main Methods:

• Conceptual modeling approach.
• Literature review on AI trust, explainability, and reliability in healthcare.
• Synthesis of existing theories and empirical evidence.

Main Results:

• A conceptual model is proposed, illustrating the pathways from AI explainability and reliability to clinician trust.
• The model posits that enhanced explainability and reliability positively influence clinician trust in AI-CDSS.
• Clinician trust is identified as a key mediator for the intention to use AI-CDSS.

Conclusions:

• Explainability and reliability are critical factors for building clinician trust in AI-CDSS.
• Trustworthy AI-CDSS design is essential for successful adoption and integration into clinical practice.
• The conceptual model serves as a foundation for future empirical research and AI-CDSS development.