A team of researchers from the University of Edinburgh says this advance in cardiac diagnostics could significantly reduce hospital admissions. The AI tool can quickly identify patients who can be sent home.
Researchers trained a new artificial intelligence algorithm named CoDE-ACS on data from 10,038 patients. All were admitted to the hospital with suspected heart attack. The AI can now make diagnoses very quickly.
Compared with current methods, the software can rule out heart attacks in twice as many patients. Its diagnostic accuracy is 99.6 percent.
What else is special about this?
Researchers say the AI tool will reduce the burden on emergency departments and reassure patients who complain of chest pain.
The gold standard for diagnosing a heart attack involves measuring troponin protein levels in the blood, which come from heart muscle cells. But factors such as age, sex, and other health conditions that affect troponin levels are often overlooked by clinicians, which can skew diagnosis. The team says their new tool accounts for those factors.
The AI takes patient information — age, sex, medical history, ECG results, and troponin levels — and analyzes it to determine the likelihood of a heart attack. It then outputs a probability score from 0 to 100, according to the Daily Mail.
Professor Nicholas Mills, who led the research, said that for patients with acute chest pain, early diagnosis and timely treatment save lives.
Final clinical trials of the AI tool are currently underway in Scotland to assess how much it can ease overcrowded emergency departments.
The results were published in the journal Nature Medicine. The research was funded by the British Heart Foundation.
Context In the United Kingdom, around 100,000 hospitalizations for heart attacks are recorded each year — about one every five minutes. Previous studies have shown that women are 50 percent more likely than men to receive an incorrect initial diagnosis. Patients who are given an incorrect diagnosis have a 70 percent higher risk of death within 30 days.
