Designing and Evaluating Trust Mechanisms in AI-Based Diagnostic Systems in Healthcare : A User-Centered Information Systems Perspective

Sammanfattning

AbstractThis thesis addresses the growing use of Artificial Intelligence (AI) in healthcare, particularly AI-based diagnostic systems designed to support healthcare-related decision-making. Although these systems offer significant potential for improving healthcare services and diagnostic processes, their successful adoption depends largely on whether users trust them. Trust in AI-based healthcare systems remains a complex issue, especially in situations involving uncertainty, risk, and limited understanding of how AI-generated outputs are produced. The study aims to explore how users perceive trust when interacting with AI-based diagnostic systems in healthcare and to examine how different trust mechanisms influence users’ willingness to rely on such systems. The research is conducted from a user-centered Information Systems perspective and focuses on both patients and healthcare professionals. The study adopts a qualitative and interpretive research approach. Data were collected through semi-structured interviews with patients and healthcare professionals who have experience using AI-based healthcare technologies. The empirical data were analyzed using thematic analysis guided by the study’s conceptual framework. The findings show that trust in AI-based diagnostic systems was conditional, dynamic, and context-dependent rather than absolute. Users generally perceived AI systems as supportive tools rather than replacements for professional expertise. Several trust mechanisms were found to influence users’ perceived trust, including transparency, explainability, usability, user experience, ethical perceptions, errors, and system limitations. The findings also highlight the importance of verification, professional expertise, and perceived risk in shaping users’ willingness to rely on AI-generated outputs. The study contributes to Information Systems research by providing a user-centered understanding of trust in AI-based diagnostic systems in healthcare. It further contributes by conceptualizing transparency, explainability, usability, user experience, and ethical perceptions as trust mechanisms influencing users’ perceived trust during interaction with AI systems. The study also highlights the importance of contextual and interactional factors in shaping trust in AI-based healthcare systems.