Artificial Intelligence (AI) was used to develop an algorithm that can assist doctors in diagnosing heart attacks in women more precisely and swiftly than ever before, according to research funded and presented at the ESC (European Society of Cardiology) Congress in Barcelona.
The previous research funded by BHF revealed that UK women with a heart attack received poor care at each stage than men. Women were 50% more likely than men to receive an incorrect initial diagnosis, emphasizing the need for innovations to assist in closing the heart attack gender gap.
The current gold standard for diagnosing a heart attack is to measure the protein troponin in the blood. The levels of troponin released by the heart, however, differ between men and women, as well as with age and other health conditions. Current guidelines apply the same threshold to all patients, which means that current tests are not as precise as they could be.
More precise than standard tests
The University of Edinburgh researchers combined Data from 10,038 people (48 percent of whom were women) who went to the hospital with a suspected heart attack to create an AI-based tool to assist clinicians to diagnose heart attacks more precisely. They then validated it on an additional 3,035 people (31% of whom were women) outside of the UK.
The tool known as CoDE-ACS uses Artificial Intelligence to combine routinely collected patient Data when they arrive at the hospital (such as gender, age, observations, ECG findings, and medical history) with the results of the troponin blood test. This yields a score between 0 and 100.
The team discovered that CoDE-ACS could rule out a heart attack with 99.5 percent accuracy, indicating that they were safe to go home. It also identified those who would benefit from remaining in the hospital for additional testing if the final analysis was a heart attack, with an accuracy of 83.7 percent. In comparison, current tests have an accuracy of only 49.4 percent. Using current approaches, less than half of those identified for further testing had a heart attack diagnosis. The tool performed consistently regardless of gender, age, or pre-existing health conditions.
Integrating into a doctor’s support app
Dimitrios Doudesis – a Data scientist at the BHF Centre for Cardiovascular Science at the University of Edinburgh, who presented the research, stated: We’ve now integrated our algorithm into a simple mobile app to assist doctors in making treatment decisions. While the troponin test takes 30 minutes to complete, we incorporate a variety of other health data into the app alongside the troponin measurement. This gives doctors a precise and instantaneous score, allowing them to determine whether they can reassure their patients that they haven’t had a heart attack and send them home, or if they need further tests.
According to BHF Professor Nick Mills of the BHF Centre for Cardiovascular Science at the University of Edinburgh: Every year, half a million people in the UK seek medical attention for chest pain, with over 200,000 seeking treatment for heart attacks. Our current approach is flawed, with one in every five people returning after 30 days and one in twenty having a heart attack or dying from cardiovascular disease after a year. Our mission is to accelerate improvements in heart attack diagnosis, and we hope that our CoDE-ACS app will be used in Emergency Departments across the UK to provide more individualized care and better outcomes.
