Health artificial intelligence (AI) is often developed in high-stakes, data-scarce contexts, where both clinical validity and patient comprehension are critical; however, rigorous, multi-level evaluation of explanations in real-world patient-facing settings remains challenging. To enhance patient understanding and trust, we propose a practical blueprint for designing and evaluating medically aligned, patient-centered explanation (MAP-X). We propose this blueprint through MAP-X, a system that employs a large language model (LLM) with retrieval-augmented generation (RAG) to translate clinical assessments into an understandable interface. We conducted a three-phase evaluation following a multi-level validation framework: a functional evaluation of faithfulness, a clinician evaluation of workflow suitability, and a patient evaluation of perceived understanding and trust. Our findings suggest that MAP-X may support clinical adoption. In the patient study, MAP-X showed higher reported trust and a positive trend in explanation satisfaction. Interviews suggested clearer understanding of assessment results. Overall, MAP-X produced clinically relevant explanations with reasonable faithfulness and usability. Clinician oversight remains necessary.

Objectives This study aimed to assess the impact of cognitive training (CT) chatbots on older adults’ memory function and mental health. Specifically, it focused on the effects of engagement and adherence.Methods: A CT chatbot developed for this study incorporates motivational interviewing strategies and personalisation features to enhance engagement and adherence. Thirty-two participants (M = 73.3 years, SD = 5.85) used it for 90 days. Measures included Delayed Matching to Sample (DMS) and Paired Associated Learning (PAL) for memory, and depression and anxiety scales. Multiple regression analysis was used to identify the influence of engagement and adherence on these outcomes.Results: Engagement and adherence statistics were significant, with the percentage of days that the participants logged into the chatbot of 86.78%. Improved memory function was observed (p = 0.01); higher engagement was associated with better DMS scores (p = 0.001) and was linked to lower anxiety levels (p = 0.033), while greater adherence was correlated with reduced depression (p = 0.014).Conclusion: This study highlights the importance of engagement and adherence in enhancing CT chatbots’ effects on memory, depression, and anxiety among older adults. These insights suggest optimising chatbot-based cognitive training should focus on strategies that improve engagement and adherence.

Explainable AI (XAI) is increasingly being used in the healthcare domain. In health management, clinicians and patients are critical stakeholders, requiring tailored XAI explanations based on their unique needs. Our study investigates the differences in explanation needs between clinicians and patients and designs corresponding explanation interfaces for each group. Using a scenario-based approach, we assessed stakeholder-tailored needs, analyzed differences, and designed interfaces using theoretical frameworks. The results demonstrate diverse stakeholder motivations for seeking explanations, leading to varied requirements. The designed interfaces effectively address these requirements, as validated by the preference selection and qualitative feedback from clinicians and patients. Their suggestions provide design insights and highlight the divergent needs of these stakeholder groups. This study contributes practical and theoretical implications to XAI research, emphasizing the importance of understanding diverse stakeholder needs and incorporating relevant theoretical concepts into user-centered interface design.