Related Experiment Video
Updated: Nov 6, 2025
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Predicting Ambulance Patient Wait Times: A Multicenter Derivation and Validation Study
Katie J Walker1, Jirayus Jiarpakdee2, Anne Loupis3
1Cabrini Emergency Department, Malvern, Melbourne, Victoria, Australia; Cabrini Institute, Malvern, Melbourne, Victoria, Australia; Casey Emergency Department, Berwick, Melbourne, Victoria, Australia; School of Clinical Sciences at Monash Health, Monash University, Clayton, Melbourne, Victoria, Australia.
Machine learning models can predict emergency ambulance patient wait times, offering valuable insights for emergency departments. These models utilize readily available data to improve operational efficiency and patient care.
Area of Science:
- Emergency Medicine
- Health Informatics
- Machine Learning Applications
Background:
- Emergency departments (EDs) face challenges in managing patient flow and wait times.
- Predicting ambulance patient off-stretcher times is crucial for resource allocation and patient care.
Purpose of the Study:
- To develop and validate machine learning (ML) models for predicting emergency ambulance patient door-to-off-stretcher wait times.
- To ensure model applicability across diverse emergency departments.
Main Methods:
- Utilized retrospective administrative data from nine Australian EDs (2017-2019).
- Developed and validated statistical and ML models, including linear regression and elastic net.
- Analyzed 421,894 patient episodes.
Main Results:
- Median off-load times varied significantly across sites (13-29 minutes).
- Global ML models achieved median absolute errors of 11.7-12.8 minutes.
- Individual site models showed higher accuracy (6.3-16.1 minutes), with key predictors including recent patient wait times, triage category, and age.
Conclusions:
- Electronic demographic and flow data can accurately estimate ambulance patient off-stretcher times.
- ML models can be built with reasonable accuracy for multiple hospitals using limited point-of-care variables.
