Perceived Trust and Professional Identity Threat in AI-Based Clinical Decision Support Systems: Scenario-Based...

Area of Science:

• Medical Informatics
• Human-Computer Interaction
• Artificial Intelligence in Healthcare

Background:

• Artificial intelligence (AI) systems, including clinical decision support systems (CDSSs), show promise in healthcare, sometimes exceeding human specialist performance.
• However, AI integration may challenge medical professionals' identities and erode trust, potentially leading to rejection of these technologies.

Purpose of the Study:

• To investigate how AI process design features influence physicians' trust in AI solutions.
• To examine the impact of AI design features on perceived threats to professional identity.
• Key features studied include AI explainability, workflow integration depth, and physician accountability for AI-driven decisions.

Main Methods:

• A 3-factorial, web-based, between-subject experiment was conducted with 292 participants (medical students and experienced physicians).
• Participants were presented with scenarios involving an AI-based CDSS for sepsis prediction and prevention.
• A 2x2x2 factorial design manipulated AI explainability, integration depth, and accountability.

Main Results:

• AI explainability positively correlated with trust and professional identity threat perceptions.
• Trust in AI was negatively associated with identity threat perceptions, indicating a partial mediation effect.
• Deep integration of AI advice into clinical workflows positively correlated with trust; accountability positively correlated with identity threat perceptions.

Conclusions:

• AI process design significantly shapes professional identity perceptions, with trust acting as a mediator.
• Explainable AI and deep workflow integration enhance trust, mitigating identity threats.
• Explainable AI and individual accountability can directly increase perceived threats, highlighting the complex behavioral impact of AI in healthcare.