A clinical trial involving 50 hospital intensive care units has been funded to test if artificial intelligence can help doctors save the lives of more patients on life support.The 'REVOLUTION trial' will use machine learning to evaluate the best targets for delivering oxygen to patients on life support.It will involve ICU's in Australia and New Zealand and is being led by the co-clinical leader of intensive care at Wellington Hospital, Professor Paul Young.He has just been awarded a $5 million grant from the Health Research Council of New Zealand for the trial and hopes that more than 24,000 patients will participate.Young told Nine to Noon the goal was to move towards providing evidence to allow clinicians to individualise care based on what works best for them individually, not just based on the general population."I guess there's been this thing in medicine where there's this tension between evidence-based medicine - that is the evidence that comes from clinical trials - and the idea that you need to personalise medicine - that is, provide the right treatment for the patient in front of you."This is really designed as a means to bridge that gap by creating personalised evidence-based medicine."Professor Paul Young.RNZ / Kate GreenYoung said the trial was a world-first and had the potential to be "paradigm-shifting".Another trial, which recently concluded, saw 40,000 patients in 135 ICUs in 15 countries randomly allocated more or less oxygen while on life support.The data collected from this trial will be used to build the machine needed for the 'REVOLUTION trial'."Essentially we will have 40,000 people where we have the baseline characteristics that describe those patients, the treatments they were allocated to and the outcomes that we observed," Young said."So then when we see a new patient, we can say 'given the set of baseline characteristics that this patient has, which treatment looks like the right one to provide this patient if we want the outcome to be alive?'."Young explained that all patients on life support - or on a breathing machine in ICU - receive an amount of oxygen that is considered safe.Some receive more, some receive less, and it is given by a nurse after a doctor prescribes it.The amount they receive can vary depending on what is wrong with the patient - and the best apporach for each person remains uncertain, Young said.The uncertainty about the optimal amount of oxygen for each patient is what has driven this project, he said.With this trial, clinicians will have a "very large scale of robust data" to apply models to.They will be able to use the calculator developed and enter the patient's information - which will give a prediction for what amount of oxugen will be best for that individual."We're aiming here to reduce the risk of death for people who require unplanned life support."Once the model is built, Young said the trial would take 22 months.Nurses and doctors will be doing the "same thing" - "just with a set of instructions".Young said they were at an "exciting stage" where the gap to bridge personalised treatments for individuals "looks like it might be able to be bridged".