Designing Explainable AI for Healthcare Reviews: Guidance on Adoption and Trust

Explainability is crucial for trust in healthcare AI: 84% of survey respondents (N=60) considered it important to understand why a review was classified by the AI system, and 82% said explanations would increase their trust—demonstrating that transparency directly influences whether patients will adopt and rely on these tools.

Patients want practical, accessible explanations: The study identified core requirements including accuracy, clarity, simplicity, and data credibility. Notably, 45% of respondents preferred combined text-and-visual explanations, suggesting that one-size-fits-all explanation formats may be insufficient for diverse audiences.

Clear communication drives adoption: 82% of respondents agreed the system highlights essential information and 82% said it saves time, but the research emphasizes that complexity and inaccuracy actively hinder adoption—indicating that even useful AI systems will fail if their outputs aren't presented clearly and reliably.

Design must be audience-aware and layered: The paper recommends "layered, audience-aware explanations" rather than uniform approaches, acknowledging that different users (patients, healthcare providers, administrators) may need different levels of technical detail and explanation styles to make informed decisions.

Expert input reveals hidden implementation challenges: Interviews with AI experts uncovered technical considerations and potential challenges for different explanation methods that survey respondents alone might not have identified, highlighting the importance of involving multiple stakeholders in responsible AI design.