REANNZ develops and supports a range of products and services to support the specialist needs of our members in the R&E and innovation community.
We operate NZ's national research and education network and seamlessly connect with 120+ networks globally to enable researchers to collaborate.
REANNZ is proud to support the specialist needs of our members from NZ's research, education and innovation community.
Find out more about who REANNZ is, what we do and the people that operate New Zealand’s national research and education network.
Researchers are using the REANNZ HPC platform to develop a machine learning algorithm that could improve ambulance dispatch services.
Ambulance response time is critical to patient outcomes in medical emergencies.
For conditions such as cardiac arrest, every minute counts. According to Wellington Free Ambulance, for every minute without CPR (cardiopulmonary resuscitation) or defibrillation, a cardiac patient’s chance of survival falls by 10–15%.
In 2024, New Zealand's median ambulance response time for urban out-of-hospital cardiac arrest events was about 9 minutes, and about 12 minutes in rural and remote communities. Similarly, every minute treatment is delayed, a stroke patient ages by 3.1 weeks. There are roughly 9000 strokes in New Zealand each year.
When responding to emergencies, every second counts. Pictured above, Wellington Free Ambulance vehicle enroute to a call. Image courtesy of Wellington Free Ambulance.
At Victoria University of Wellington - Te Herenga Waka, PhD student Jordan MacLachlan and his Supervisor A/Prof. Yi Mei are developing a machine learning algorithm to provide dispatch recommendations for ambulance services.
The model considers real-time factors such as traffic, travel time, and the locations of other ambulances to improve operational efficiency such as to minimise both paramedic workload and emergency response times. His project required training thousands of models on millions of days of emergency data.
This kind of scale demands substantial computational power and efficient algorithms.
"Our work, funded by an MBIE Smart Idea grant, would not have been possible without access to national eResearch infrastructure," Jordan said.
By integrating artificial intelligence techniques, Jordan’s algorithm automatically evolves a high performing model. Each of the seven dispatch decisions are represented by a “priority tree” that evaluates factors such as distance, urgency, and availability.
Over successive iterations, the best-performing models are combined and refined to improve overall system performance, much like natural selection in biology.
His resulting model has consistently outperformed human dispatch decisions in simulations using historic data, demonstrating faster response times as a result of making more efficient resource allocation decisions.
The effectiveness of this approach has recently been validated by a third party over one year of historic data. Jordan’s technology was shown to reduce the mean response time to cardiac arrests by 46 seconds, to improve the mean (and 90th percentile) response times to all urgency classes, and to improve all paramedic break KPIs.
The impact of this technology has been described by those in the ambulance industry as “a once in a generation shift for ambulance services.”
As an additional information and decision-making tool, this model could help emergency services shave valuable seconds off response times, potentially saving more lives.
Progressing this project from concept to a high-performing model wouldn't have been possible without access to the computing resources and expertise provided by REANNZ.
"The HPC Platform and support team were invaluable, and we look forward to working with REANNZ Research Software Engineers to further optimise our workflow," Jordan said.
Find anything about our products, services, and more. Enter a query in the search input above.
