$5 million to explore using AI to guide oxygen therapy decisions in ICU

NEWS  - eHealthNews editor Rebecca McBeth

A major clinical trial involving 50 hospital intensive care units (ICUs) across New Zealand and Australia will test whether artificial intelligence (AI) can help doctors improve survival for patients on life support.

Paul Young, deputy director of the Medical Research Institute of New Zealand and co-clinical leader at Wellington Hospital ICU, has received a $5 million grant from the Health Research Council (HRC) to lead the the Randomised EValuation of Oxygen Limits Using Tailored IndividualisatiON (REVOLUTION) trial.

He says that hundreds of thousands of lives could be saved each year in ICUs around the world, even if the trial confirms only a modest reduction in deaths

The trial will recruit more than 24,000 patients and is the first of its kind worldwide to evaluate whether AI-guided treatment using machine learning improves survival in the ICU. 

A HRC statement says that if successful, it could set an international benchmark for the safe and rigorous application of AI-derived models in clinical decision-making.

The grant is one of two programme grants with a combined value of $10 million and 38 project grants, with a combined value of $46.8 million, announced by the HRC. 

Clinical trials are a key government priority and this year's funding is supporting eight across the country, including $1.5 million for a project called SENSAI which will investigate a smartwatch early notification system for asthma intervention.

Young explains that oxygen is the most widely used therapy in intensive care, but there is growing evidence that the optimal oxygen level differs between patients.

The aim of the trial is to use machine learning to “transform critical care from a 'one-size-fits-all' model to a personalised care model.

“If successful, this research could fundamentally change how oxygen therapy is delivered in intensive care units around the world, improving outcomes for thousands of critically ill patients each year while demonstrating how artificial intelligence can be safely and effectively integrated into routine clinical care,” he tells eHealthNews. 

“It could also pave the way to use this method to personalise treatments in other areas of medicine.”

Young says they will first refine and optimise their machine learning model using data from their recently completed HRC-funded Mega-ROX trial, the world’s largest ICU trial involving 40,003 patients from 137 ICUs in 14 countries. 

The model will use these data, together with personal data submitted by clinicians when patients are admitted to the ICU, to estimate the individual benefit, or harm, of higher versus lower oxygen levels for patients on life support.

From there, they will compare the personalised delivery of oxygen treatments to patients when doctors use machine learning to help them decide what level of oxygen each patient receives, versus the standard approach when doctors decide the oxygen level without AI.

Young says a limitation of trials like Mega-ROX is that they measure average treatment effects across a population. 

“Applying those averages to individual patients assumes uniform benefit, an assumption that may not reflect clinical reality,” he says.

REVOLUTION will use machine learning to identify the oxygen target most likely to benefit each individual patient.

Of the 50 ICUs involved, 12 are based in New Zealand and are expected to contribute around 7,000 patients. 

Young says funding has also been secured in Australia to recruit the remaining patients from Australian ICUs. The trial is expected to begin recruiting in 2028, with participants being adults who require unplanned life support in the ICU.

Young says the team is delighted and relieved to get the HRC funding.

"We have a highly skilled research co-ordinator workforce whose employment depends on grants like these," says Professor Young. 

"As funding levels for health research have reduced in recent years, it has become increasingly challenging to get even the very best projects funded."

HRC director of investments and co-chief executive (acting) Stacey Pene says the trial bridges the gap between research and clinical practice at a national level.

"This study has the potential to fundamentally change the way clinical trial results are translated into patient care by integrating machine learning with high-quality data from a randomised clinical trial," he says.

Image: Paul Young, deputy director of the Medical Research Institute of New Zealand and co-clinical leader at Wellington Hospital ICU 

If you would like to provide feedback on this news story, please contact the editor Rebecca McBeth.

You’ve read this article for free, but good journalism takes time and resource to produce. Please consider supporting eHealthNews by becoming a member of HiNZ, for just $17 a month.

Read more AI & Analytics news

Return to eHealthNews.nz home page