Equipping AI Chatbots to Address Patient Distress and Suicidal Thoughts in High-Risk Healthcare Environments
By
Joanna M. Streck
Dallas Swendeman
W. Scott Comulada
Y. Xian Ho
Julia Cannistraro
Gladys N. Pachas
Danielle Alves-Back
Wei Sum Li
Jonathan L. Jackson
Delta-Marie Lewis
Namrata Tomar
David Warren
Douglas E. Levy
Lillian Gelberg
September 22, 2026
Clinical Scorecard: Equipping AI Chatbots to Address Patient Distress and Suicidal Thoughts in High-Risk Healthcare Environments
At a Glance
Category Detail
Condition Behavioral Health Support Key Mechanisms Generative AI chatbots utilizing large language models for real-time dialogue and support. Target Population Adults with substance use disorders (SUD), including opioid use disorder (OUD). Care Setting Primary care-based addiction treatment clinics.
Key Highlights
AI chatbots can provide behavioral health support when human clinicians are unavailable. Suzy, an AI chatbot, was developed to assist SUD recovery and monitor suicide risk. Safety guardrails and monitoring procedures are essential for AI chatbot deployment. Iterative design refinement involved expert review to enhance risk assessment capabilities. AI chatbots may help address gaps in access to mental healthcare. Guideline-Based Recommendations
Diagnosis
Management
Implement safety disclaimers and triage systems for suicide risk monitoring. Monitoring & Follow-up
Utilize a 'Safety Router' to assess and triage user messages based on risk levels. Risks
AI chatbots may provide inaccurate or harmful advice and lack human oversight. Patient & Prescribing Data
Adults receiving medication treatment for opioid use disorder.
AI chatbots can enhance support but require rigorous safety monitoring.
Clinical Best Practices
Incorporate direct feedback from clinicians in the design of AI chatbots. Establish clear protocols for emergency referrals when users express suicidal thoughts. Related Resources & Content
