Huimin Liu

The perinatal period can leave women vulnerable to mental health disorders, and artificial intelligence (AI) holds potential to enhance prediction, diagnosis, and treatment applications, yet its clinical integration in perinatal mental health remains underexplored. Using a qualitative design supplemented by exploratory quantitative analysis of theme frequencies. Grounded in Shenzhen’s three-tier perinatal mental health prevention and intervention system, this study examined the ethical expectations, concerns, and implementation requirements for AI in perinatal mental health through semi-structured interviews with 41 participants, including 23 maternal health workers (MHWs) and 18 perinatal women. Findings revealed ethical divergences: MHWs prioritized clinical validity, diagnostic reliability, and system feasibility, whereas perinatal women emphasized data privacy, personalized care, and equitable access. Across both stakeholder groups, participants agreed that AI should serve as a supportive tool to enhance assessment, complementing rather than replacing human relational care. This study outlines a dual-stakeholder framework for AI integration in perinatal mental health, structured around four derived pillars: intelligent assessment, equity, tool governance, and clinical validation, each grounded in distinct priorities of health workers and patients.