How Can Explainable AI Improve Trust and Transparency in Medical Diagnosis Systems

Abstract

The active adoption of AI in human services has provoked the problem of transparency and trust in decisions, since most healthcare AI systems are black-box models. To resolve these concerns, explainable Artificial Intelligence (XAI) has been suggested as a means of making medical AI tools safer, more reliable, and acceptable through human-readable explanations. This research paper describes the role of XAI in physician trust and acceptance of AI-based diagnostics. The knowledge of XAI, AI decision confidence, perceived usefulness, and adoption intentions were some of the main variables evaluated in a structured survey of 30 medical students and an expert interview. The findings indicate that participants unanimously concurred that AI explanations raise the clarity, safety and acceptability of AI recommendations. The knowledge of XAI showed a strong positive relationship with trust (r = 0.48, p = 0.01) and perceived usefulness (r= 0.60, p = 0.001). The model has demonstrated a steady reliability level (Cronbach a = 0.702) and accounted 48-52% variance. Although the study has some limitations, including the small size of the sample and self-reporting, it still has empirical evidence of the positive effect of XAI on human-AI collaboration on the necessary condition of successful integration of AI diagnostic tools in healthcare. Further studies need to be conducted in a clinical inquiry of XAI and determine institutional and patient attitudes.

Ähnliche Arbeiten