Doctors could soon be using artificial intelligence in diagnosing heart attacks, according to New Zealand-led research that provides a glimpse into of the future of healthcare.
The study, published today in journal Circulation, found using an algorithm combined with blood test results could provide a faster and more accurate diagnosis of patients presenting with heart attack symptoms.
"This will facilitate us to provide early treatments to those that need it and we would reduce wasted time and resources for those who do not have heart attacks," study author Dr Martin Than, of Christchurch Hospital, said.
Public awareness campaigns were doing a great job of highlighting heart attack symptoms, which included chest pains and dizziness, among others, Than said.
But this meant a lot of people presenting at emergency departments were not actually experiencing a heart attack, which put strains on health systems.
In the United States about 8 million people appeared at hospitals each year with heart attack symptoms, with about 85 per cent not experiencing a heart attack.
In New Zealand a similar proportion of the 50,000 patients presenting with such symptoms each year were not having a heart attack.
"This runs the risk of holding a patient in hospital longer than necessary, which can put stress on the patient and also use up resources that could be used for people having actual heart attacks," Than said.
Currently, people going into hospital with a suspected heart attack were assessed by a doctor and given an electrocardiogram (ECG) and a troponin blood test.
This test essentially gave the patient score, which doctors would then combine with various factors - including age and sex - to determine the likelihood of them having a heart attack.
The algorithm, developed by healthcare company Abbott, combined a person's characteristics with the blood test results to give doctors a faster and more accurate heart attack diagnosis, or the ability to more confidently discharge them from the hospital.
"We have taken all of that information and assimilated it into a huge data set so we can say rather than a single cut off point, we can give that exact person a percentage tailored to them," Than said.
The algorithm was one of the first effective demonstrations of how machine learning - a subset of artificial intelligence - could be used to guide clinical decision-making in patients suspected of having a heart attack.
"This algorithm improves the speed, accuracy and personalisation of data analysis and removes the bias and potential inaccuracy that may happen with human calculations and the one-size-fits-all approach. It will give patients a visual display, which they can then use to make a decision with their doctor."
The algorithm could also predict the risk of a heart attack over the following 30 days.
There was always a risk of misdiagnosis, Than said.
"The only way to rule that out is to keep everybody in hospital. The challenge with this is to keep the same level of safety as currently, but diagnose in a manner far more accurate and earlier."
Than said while there was no set timeline for when this technology could be introduced to hospitals here, there were trials planned around the world.
"It is very exciting. In 10 years time this technology could be used for many important things people come into hospital for, not just heart attacks. In the future it will be far more precise."
The international study involved 11,011 patients from nine countries, including New Zealand.