Scientists from the University of Edinburgh found that artificial intelligence can rule out the presence of a heart attack in emergency patients with 99.6% accuracy. Research published in the journal nature medicine.
The CoDE-ACS algorithm was developed using data from 10,038 patients hospitalized for suspected heart attack in Scotland. It uses widely collected information about the patient: age, gender, EKG results, medical history, and troponin levels.
The algorithm’s effectiveness was tested on 10,286 patients in six countries around the world. The researchers found that, compared with existing testing methods, CoDE-ACS was able to rule out heart attacks with 99.6% accuracy in twice as many patients over the same time period.
Currently, the gold standard for diagnosing heart attack is measurement of troponin protein levels in the blood. However, the same threshold is used for each patient. This means that factors such as age, gender, and other health problems that affect troponin levels are not taken into account and affect the accuracy of a heart attack diagnosis. This may lead to inequalities in diagnosis.
For example, a previous study found that women were 50% more likely to receive a false initial diagnosis and therefore had a 70% greater risk of death in the next 30 days. CoDE-ACS was effective regardless of age, gender, or underlying medical conditions. The accuracy of the new algorithm did not depend on the gender or age of the patients.