Intelligent Patient Triage System

The Challenge

MedPriority's emergency departments were overwhelmed. Patient volumes had increased 35% over two years, but staffing hadn't kept pace. Triage nurses were making rapid assessments under pressure, and consistency varied significantly between shifts and locations.

The consequences were real: undertriaged patients waited too long for critical care, while overtriaged patients consumed resources needed elsewhere. With three clinic locations and different nurse experience levels, they needed a system that could bring consistency without replacing clinical judgment.

Our Approach

We worked closely with MedPriority's senior physicians to codify their triage decision-making into a structured framework. Then we built an AI assistant that analyzes patient symptoms, vitals, and medical history to recommend a triage priority level — with the nurse always making the final call.

The model was trained on 24 months of triage records cross-referenced with patient outcomes, so it learned to identify patterns that predict acuity — not just match symptoms to textbook categories.

The Solution

A mobile app built for the triage workflow: nurses input symptoms and vitals (or scan from connected devices), and the AI provides a recommended priority level with supporting reasoning in under 3 seconds. The system integrates with MedPriority's existing EHR via HL7 FHIR APIs.

We built a confidence indicator — when the AI is uncertain, it explicitly flags the case for senior review rather than guessing. This was essential for clinical trust and patient safety.

Results