AI-powered patient triage system
HealthcareAIHealthcareAutomation2025

HealthCore

AI-powered patient triage system

4,000+ / yr

Clinical hours recovered

↓ 61%

Average triage time

94.3%

Model accuracy

14 / 14

Sites live

The challenge

Manual triage across 14 hospital sites created dangerous bottlenecks and inconsistent prioritisation during peak periods.

HealthCore operates 14 acute care hospitals across three regions. During peak admissions — winter months, public holidays, major incidents — their manual triage process created life-critical delays. Nurses were spending up to 40% of their shift on administrative prioritisation rather than patient care. Maple & Chen was engaged to design, build, and deploy an AI triage system that could assess patient urgency at point of registration and route cases to the right care pathway automatically.

Our approach

1

Clinical data architecture

We began with a six-week discovery embedded across three hospitals, interviewing 40+ clinicians and reviewing 18 months of anonymised case data. The first task was establishing a clean, standardised data model — existing records were inconsistent across sites, with different codes for the same conditions.

2

Model development & validation

A custom multi-class classification model was trained on 220,000 historical triage decisions, validated against clinical outcomes and reviewed by HealthCore's chief medical officer before any patient contact. The model outputs a urgency score with a confidence interval — below a confidence threshold, cases are flagged for immediate human review.

3

Integration without disruption

The system was integrated into HealthCore's existing EPR via a middleware layer — no rip-and-replace of clinical systems. Rollout was staggered across sites over 12 weeks, with a parallel-run period at each site so clinical leads could validate outputs before full go-live.

4

Ongoing monitoring

A live monitoring dashboard tracks model accuracy, drift, and edge-case flags daily. HealthCore's IT team was trained to manage the pipeline; Maple & Chen holds a support retainer for model retraining on a quarterly cycle.

We'd been told for years that AI in clinical triage wasn't viable without massive infrastructure change. Maple & Chen proved that wrong — and did it without disrupting a single shift.

Dr. Sarah Omotunde

Chief Medical Officer, HealthCore

Ready to move forward?

Let's navigate complexity together.

Every great partnership begins with a conversation. Tell us what you're building — we'll show you how to get there.

Begin a conversation