AI Could Help Narrow Heart Attack Gender Gap
According to research being presented in Barcelona at the European Society of Cardiology Congress, doctors will be able to diagnose heart attacks amongst women quicker and more accurately than before.
This would be because of an algorithm that has been developed with the use of artificial intelligence.
The need for care
Previous research funded by BHF has shown that women who get a heart attack in the United Kingdom don’t receive the same level of care at every stage, as opposed to men.
There is a 50% more chance of women getting a wrong initial diagnosis, which highlights the need for innovations that can help in closing the gender gap in heart attacks.
The current method used for a heart attack diagnosis is to measure the levels of the protein troponin in the blood.
But, the levels of troponin that are released by the heart are different for men and women and they also depend on other health conditions and age.
The same threshold is used for the diagnosis of all patients under the current guidelines, which means that the accuracy of the tests is not as good as it could be.
Greater accuracy than standard tests
The University of Edinburgh’s researchers took the data of about 10,038 people who suspected they had a heart attack and visited the hospital. 48% of these patients were women.
They used the data for developing a tool based on artificial intelligence (AI) for helping clinicians in giving a more accurate diagnosis of heart attacks.
The tool was then further validated on 3,035 people outside of the UK, 31% of whom were women.
The tool has been named CoDE-ACS and it uses artificial intelligence for combining the results of their troponin blood test with the information routinely collected when they come to a hospital.
This includes information, such as age, sex, medical history, ECG findings, and observations. Then, a score is produced between 0 and 100.
The team discovered that the tool was 99.5% accurate in ruling out a heart attack and confirmed that the patients could go home safely.
It was also able to identify which patients could benefit from staying at the hospital for more tests and the final heart attack diagnosis was made in these patients with 83.7% accuracy.
Current tests only have an accuracy of 49.4%. The tool was able to deliver consistent performance, regardless of age, gender, or any pre-existing health conditions.
Level playing field
The Associate Medical Director, Professor James Leiper said that this was a major step forward that would help in leveling the playing field for diagnosing and treating a heart attack.
He said that they were aware that chances of misdiagnosis were higher in women, but this team could provide a solution that would eliminate the problem.
The problem with current tests is that the troponin levels of some patients do not fit the threshold for ruling in or out, which can make clinical decisions challenging.
But, the new tool was able to refine the risk in about 29.5% of patients who did not fit with the rule in and out criteria, allowing for more accurate results.
