AI Triage Assistant for a Multi-Specialty Hospital Chain
The Challenge
Emergency department triage was a serious bottleneck. During peak hours, patients waited 45 minutes or longer before an initial assessment — a delay that affected both patient outcomes and satisfaction scores. The nursing staff was stretched thin, spending significant time on structured intake conversations that followed predictable protocols regardless of patient complexity.
Our Approach
Tequity built a real-time voice AI that conducts structured triage interviews with arriving patients using a conversational interface deployed on a tablet kiosk at registration. The AI asks clinically validated triage questions, dynamically adjusting based on responses to identify red-flag symptoms. It assesses acuity using the Emergency Severity Index framework and routes patients to the appropriate care level — from immediate resuscitation to fast-track.
The system was trained on the hospital's historical triage records and validated against nurse-assessed triage before going live. All interactions are logged and available to clinicians in the EMR for full audit and override capability.
The Results
Average ER wait time fell 35% in the first month after deployment. Triage throughput quadrupled, with the AI handling multiple patients simultaneously. A blind accuracy study showed 92% concordance with gold-standard nurse-assessed triage scores — exceeding the 85% target set at project kickoff. Nursing staff redirected time to clinical care rather than intake administration.
