From Waitlists to Well-Being: An Analytical Approach to AI-Supported Mental Healthcare

Abstract: The growing burden of mental health disorders is straining healthcare systems, leading to long waitlists and delayed access to therapy. Such delays can worsen symptoms, increase dropout rates, and impose substantial societal costs. Advances in Artificial Intelligence (AI)—including supervised conversational agents—offer scalable opportunities to support patients during these waiting periods. We develop a multi-class, multi-server queueing model to jointly optimize waitlist treatment and scheduling policies. Using a fluid approximation, we derive a simple, implementable policy that incorporates fixed and variable costs while capturing key behavioral features of psychotherapy, including no-shows, dropouts, and heterogeneous treatment effects. Numerical experiments show that the proposed policy closely matches the exact MDP solution in small systems, remains robust across system scales and service-time distributions, and consistently outperforms benchmark policies. We further quantify how hiring costs affects waitlist, scheduling and recruitment decisions. A case study calibrated to a representative Veterans Health Administration (VHA) mental health unit demonstrates that supervised waitlist treatment delivers substantial performance gains and, when coordinated with scheduling decisions, can significantly reduce the need for additional therapist recruitment. Overall, our results highlight the potential of analytically guided, AI-supported policies to improve access to mental health care.