Multi-agent AI Systems for Adaptive, Culturally-Concordant Care Routing in Postpartum Depression Across Medicaid-Dependent Populations

Postpartum depression (PPD) remains a critical, yet under-addressed mental health crisis, particularly among Medicaid-dependent populations who face significant barriers to culturally-competent care. Fragmented referral pathways, linguistic mismatches, and insufficient resource tailoring contribute to disparities in timely diagnosis, intervention, and recovery outcomes. This research proposes the design and deployment of multi-agent artificial intelligence (AI) systems to enable adaptive, culturally-concordant care routing for postpartum individuals at risk of or experiencing depression. By integrating reinforcement learning, knowledge graphs, and federated learning frameworks, the system dynamically personalizes care pathways based on evolving patient needs, socio-cultural profiles, health literacy, and local resource availability. Each AI agent specializes in distinct functions: risk stratification, cultural adaptation, language translation, and provider matching. Through real-time collaboration and feedback loops, agents optimize longitudinal care engagement, minimize attrition, and support equitable mental health outcomes. Data inputs include electronic health records, social determinants of health, patient-reported outcomes, and community asset databases, ensuring a holistic, non-reductionist understanding of patient contexts. Importantly, federated learning preserves data privacy across different healthcare providers and community organizations, mitigating algorithmic bias and reinforcing patient trust. Pilot simulations using Medicaid datasets demonstrate the model’s superiority in reducing missed follow-up rates and increasing culturally-appropriate referrals compared to standard rule-based triaging. This work advances AI-driven healthcare equity and positions adaptive, culturally-intelligent multi-agent systems as pivotal tools in addressing postpartum depression disparities at population scale. Future directions include real-world clinical trials, incorporation of indigenous and minority care practices, and dynamic updating of resource networks as community services evolve.