The method could help to relieve pressure on hospital emergency departments. This is because the personalised Artificial Intelligence in Myocardial Infarction Study (ARTEMIS) algorithm can be used in the outpatient and pre-clinical settings, independent of large care structures.
Chest pain is a major symptom of heart attack and the most common cause of admission to hospital emergency departments worldwide. However, only five to 25 percent of these symptomatic patients actually have an acute myocardial infarction that requires immediate treatment. To detect or rule out an acute myocardial infarction, the highly sensitive troponin level is measured in the patient's blood. Troponin is a protein complex found only in the heart muscle that is released into the blood when muscle cells are damaged. International guidelines recommend laboratory-based troponin tests to diagnose a heart attack. These tests take up to 60 minutes to interpret in the laboratory and do not take into account individual patient information such as age and gender. New rapid troponin tests, known as point-of-care (POC) tests, can measure troponin levels in about eight minutes.
The researchers demonstrated that these rapid tests can be used to accurately and efficiently diagnose heart attacks when embedded in a personalised AI algorithm. In this case, a single rapid troponin test was even superior to the recommended standard diagnostic procedure. The study used data from more than 2,500 patients in the US and Australia.
Our results show that the algorithm can rule out myocardial infarction more than twice as quickly in more than twice as many patients (about 35 percent) as compared to the diagnostic procedures recommended by conventional guidelines (about 14-15 percent), while maintaining a consistently high level of confidence of almost 100 percent," says first author Dr Betül Toprak, Department of Cardiology, University Heart and Vascular Centre, UKE. The ARTEMIS algorithm also reliably rules out myocardial infarction in about 20 percent of patients with early-onset chest pain who previously would have required a second troponin measurement after one or two hours.
Applicability outside the clinic
In the future, the use of AI in combination with the rapid test may help to reduce the burden on hospital emergency departments. Patients with a low risk of myocardial infarction could be reliably identified in preclinical, outpatient or geographically isolated care areas and would not have to be referred to emergency care in a chest pain unit,' says study leader Prof Dr Stefan Blankenberg, Director of the Department of Cardiology and Medical Director of the University Heart and Vascular Centre at the UKE.
Original publication: Betül Toprak et al., Diagnostic accuracy of a machine learning algorithm using point-of-care high-sensitivity cardiac troponin I for rapid rule out of myocardial infarction: a retrospective study, Lancet Digital Health, 2024. DOI: https://doi.org/10.1016/S2589-7500(24)00191-2
Source: Press release University Medical Center Hamburg-Eppendorf (UKE)
