Home › Operations & systems › AI triage tool predicts ED admission

Operations & systems

AI triage tool predicts ED admission

The Sydney Triage to Admission Risk Tool With Artificial Intelligence (START-AI) to Support Decision Making in Emergency Departments: Model Explainability and Feature Importance Analysis
Emergency Medicine Australasia · Added 4 August 2026

Model explainability analysis of START-AI — AUROC improved from 0.78 to 0.90 with AI features, driven primarily by blood test results, triage comments, and clinical notes

Study type

Other

Design

Model explainability analysis of single-centre ED electronic medical record data over 2 years

Participants

Not reported

Main finding

AUROC for predicting inpatient admission improved from 0.78 (START alone) to 0.90 with all START-AI features; blood test results (lactate, CRP), triage comments and ED case notes contributed most, vital signs contributed little

Caveat

Single-centre data; prediction of admission rather than patient-centred outcomes

Bottom line

Context only

Learning goal

Understand how artificial intelligence triage tools using clinical notes and pathology data could support patient flow decisions in Australian emergency care.

More

• Topic: Operations & systems
• Daily scan: 4 August 2026

Related papers on Operations & systems

• Independent EMS beats hospital-based model in Chinese OHCA

  Resuscitation Plus · 9 Aug 2026 · Registry · Context only

  Retrospective analysis of 24,814 OHCA patients comparing independent EMS models with hospital-based models in China — the independent model had higher survived event rates (2.23% vs 1.40%) despite longer response times.

• NEMSIS records mostly unique encounters

  Prehospital Emergency Care · 4 Aug 2026 · Cross-sectional · Context only

  Retrospective analysis of 36.7 million NEMSIS records — 99.5% represent unique patient-care episodes with only 0.5% involving multiple records for the same encounter

• Team leaders carry highest cognitive load

  Resuscitation · 3 Aug 2026 · Cross-sectional · Context only

  This multicentre cross-sectional study of 58 resuscitations found perceived cognitive load was highest among team leaders (mean 5.5/9) and higher team leader cognitive load was associated with higher load among team members and lower TEAM-assessed performance.

• Rural vs urban EMS: 2024 NEMSIS update

  Prehospital Emergency Care · 31 Jul 2026 · Cross-sectional · Context only

  Survey examining EMS characteristics and care differences between rural and urban areas — reports on resource availability, response times, and clinical capabilities.