"Navigating the Trade-off between Explainability and Privacy in Connected Healthcare with Explainable AI"

"The integration of Artificial Intelligence (AI) into consumer wearable devices promises to revolutionize preventative medicine, yet the \"black box\" nature of these algorithms frequently undermines user trust. While Explainable AI (XAI) is proposed as a solution, its implementation introduces a critical ethical tension: the trade-off between the need for transparent health insights and the privacy risks associated with revealing granular behavioral data. This study employs a sequential exploratory mixed-methods design to investigate this \"Privacy-Transparency Paradox\" from a consumer-centric perspective. First, in-depth interviews (n=16) and scenario simulations identified that while XAI significantly enhances trust and clinical compliance, it simultaneously heightens users' sense of surveillance, creating a demand for \"honest surveillance\" over ambiguous data practices. Second, a large-scale quantitative survey (n=224) validated these findings, revealing that \"Ethical and Accountability Beliefs\" are the strongest predictor of users' intention to continue using the device, surpassing both trust and perceived value. Furthermore, results indicate that XAI serves as a crucial trust-repair mechanism in the event of algorithic failure. However, this effect is moderated by privacy cyni :: n, particularly among younger, experienced users who are more sensitive to data intrusion. This paper contributes a nuanced theoretical framework for ethical XAI in connected healthcare, arguing that future systems must move beyond binary consent models to offer tiered, context-aware explainability that balances the right to know with the right to privacy."

将人工智能（Artificial Intelligence，AI）集成至消费级可穿戴设备中，有望彻底变革预防性医疗领域，但这类算法的“黑箱”特性往往会削弱用户信任。尽管可解释人工智能（Explainable AI，XAI）被视为解决该问题的方案，但其落地却引发了一项关键伦理矛盾：即透明化健康洞察的需求与泄露精细化行为数据所带来的隐私风险之间的权衡。本研究采用序贯探索性混合方法设计，从以消费者为中心的视角探究这一“隐私-透明度悖论”。首先，通过深度访谈（样本量n=16）与情景模拟研究发现，尽管可解释人工智能可显著提升用户信任度与临床依从性，但同时也会加剧用户的被监控感，进而催生了针对模糊数据处理行为的“坦诚式监控”需求。其次，大规模定量调查（样本量n=224）验证了上述研究结果，表明“伦理与问责信念”是影响用户持续使用设备意愿的最强预测因子，其影响力超越了用户信任与感知价值。此外，研究结果显示，在遭遇算法故障时，可解释人工智能可作为一种关键的信任修复机制。不过，该效应会受到隐私怀疑主义的调节，尤其在那些对数据侵入更为敏感的年轻、有经验的用户群体中表现更为明显。本研究为互联医疗领域中符合伦理的可解释人工智能构建了一套精细化的理论框架，提出未来的系统需跳出二元同意模式，提供分层、情境感知的可解释性机制，在知情权与隐私权之间达成平衡。