Machine Learning Model Improves Diagnosis of Myocardial Infarction

THURSDAY, June 1, 2023 (HealthDay News) -- A machine learning model, incorporating cardiac troponin concentrations with clinical features, can improve the diagnosis of myocardial infarction, according to a study published online May 11 in Nature Medicine.

Dimitrios Doudesis, Ph.D., from the University of Edinburgh in the United Kingdom, and colleagues developed machine learning models that integrate cardiac troponin concentrations at presentation or on serial testing with clinical features and computed the Collaboration for the Diagnosis and Evaluation of Acute Coronary Syndrome (CoDE-ACS) score, which identifies the probability of myocardial infarction. The models were trained on data from 10,038 patients and were validated externally using data from 10,286 patients from seven cohorts.

The researchers found that CoDE-ACS had excellent discrimination for myocardial infarction (area under the curve, 0.953), with good performance across subgroups. Compared with fixed cardiac troponin thresholds, CoDE-ACS identified more patients as low probability of having myocardial infarction at presentation (61 versus 27 percent), with similar negative predictive value, and identified fewer patients as high probability of myocardial infarction (10 versus 16 percent), with greater positive predictive value. Compared with those with intermediate or high probability, those identified as having a low probability of myocardial infarction had a lower rate of cardiac death at 30 days (0.1 versus 0.5 and 1.8 percent) and one year (0.3 versus 2.8 and 4.2 percent) from presentation.

"If adopted in practice, CoDE-ACS could reduce time spent in emergency departments, prevent unnecessary hospital admissions, and improve the early treatment of myocardial infarction, with benefits for both patients and health care providers," the authors write.

Several authors disclosed ties to the biopharmaceutical industry.

Abstract/Full Text