AI-assisted health management and older adults’ autonomy: an empirical ethics study of privacy, algorithmic reliance, and relational dynamics

BACKGROUND: AI-assisted health management is increasingly integrated into healthcare practices for older adults, offering new possibilities for continuous monitoring and personalized care. However, these technologies also raise significant ethical concerns, particularly regarding privacy, autonomy, and decision-making authority. Older adults may be especially vulnerable to these challenges due to digital literacy barriers and reliance on relational support in healthcare decision-making. Empirical evidence on how these factors interact to shape older adults' autonomy in real-world contexts remains limited. METHODS: This study adopted an empirical ethics approach using a mixed-methods design, combining a cross-sectional survey with semi-structured interviews. A cross-sectional questionnaire survey was conducted among 312 community-dwelling older adults with experience using AI-assisted health management tools, followed by semi-structured interviews with a purposive subsample ([Formula: see text]). Quantitative data examined privacy awareness, algorithmic reliance, relational factors, and perceived autonomy, while qualitative interviews explored older adults' lived ethical experiences and interpretations of AI-mediated health decisions. RESULTS: Quantitative findings indicated moderate levels of privacy awareness and perceived autonomy, alongside relatively high reliance on AI-generated health recommendations. Greater algorithmic reliance was negatively associated with perceived autonomy ([Formula: see text]), and this relationship was significantly moderated by relational factors ([Formula: see text]).Qualitative analysis revealed that limited understanding of data practices often resulted in procedural rather than substantive consent, and that AI recommendations were frequently perceived as authoritative. Family involvement and healthcare professional dominance played a critical role in shaping autonomy, either enabling shared decision-making or reinforcing paternalistic decision-making structures. CONCLUSIONS: AI-assisted health management influences older adults' autonomy through intertwined ethical mechanisms involving privacy awareness, algorithmic reliance, and relational dynamics. These findings highlight the importance of relationally informed ethical frameworks and context-sensitive governance approaches to support older adults' autonomy and dignity in AI-assisted health management. CLINICAL TRIAL NUMBER: Not applicable.