New AI technology can help diagnose heart attacks


An algorithm combining high sensitive troponin testing with personal details can help A&E doctors better determine whether patients are having a heart attack, according to new research. The study, published in medical journal Circulation today, used Abbott's algorithm on 11,000 patients from the UK, the US, Germany, Switzerland, Australia and New Zealand, to see whether it could help deliver faster and more accurate evidence as to whether patients were suffering from a heart attack. Developed using machine learning – a branch of artificial intelligence – the algorithm uses a high sensitivity troponin-I blood test, and the time it was taken, to assess the patient's blood troponin protein levels, combining the results with personal details, such as age and sex, to deliver a bespoke assessment. It is thought this will help get around two current obstacles in heart attack diagnoses. The first is that women are currently at greater risk of misdiagnosis, because their troponin protein levels can be lower than those of men, and international guidelines for the use high sensitive troponin tests do not always account for sex in results.