Artificial Intelligence for Safer Crisis Response: De-escalation Prompts and Emergency Handoff Optimization

Behavioral crises in autistic patients are also a significant source of challenges for emergency departments (EDs) and emergency medical services (EMS), with autistic patients disproportionately restrained and left overnight. These interventions are associated with the potential for physical harm, psychological trauma and re-crisis, and call for safer methods based on the best available evidence. The 988 Crisis and ED Triage Copilot is an artificial intelligence (AI)-powered framework that provides real time de-escalation prompts to EMS and ED personnel while standardizing patient handoffs, as introduced in this paper. Based on the Joint Commission's patient safety goals and the Substance Abuse and Mental Health Services Administration's 988 framework, the Copilot combines natural language processing (NLP), predictive analytics and structured communication pathways to eliminate harm. Simulation results for 500 patient encounters show: reductions in use of restraint (28.8%), reduction in length of stay (LOS) by 1.4 hours, reduction in staff incident reporting by 20%. Conclusion: These data indicate that AI copilots can serve to improve safety and equity in emergency care for autistic patients while complying with national standards of crisis response and patient safety.