Machine learning for diagnosis of myocardial infarction using cardiac troponin concentrations

Doudesis, Dimitrios; Lee, Kuan Ken; Boeddinghaus, Jasper; Bularga, Anda; Ferry, Amy V.; Tuck, Chris; Lowry, Matthew T.H.; Lopez-Ayala, Pedro; Nestelberger, Thomas; Koechlin, Luca; Bernabeu, Miguel O.; Neubeck, Lis; Anand, Atul; Schulz, Karen; Apple, Fred S.; Parsonage, William; Greenslade, Jaimi H.; Cullen, Louise; Pickering, John W.; Than, Martin P.; Gray, Alasdair; Mueller, Christian; Mills, Nicholas L.; CoDE-ACS Investigators

Abstract

Although guidelines recommend fixed cardiac troponin thresholds for the diagnosis of myocardial infarction, troponin concentrations are influenced by age, sex, comorbidities and time from symptom onset. To improve diagnosis, we developed machine learning models that integrate cardiac troponin concentrations at presentation or on serial testing with clinical features and compute the CoDE-ACS score (0-100) that corresponds to an individual’s probability of myocardial infarction. The models were trained on data from 10,038 patients (48% women) and their performance was externally validated using data from 10,286 patients (35% women) from seven cohorts. CoDE-ACS had excellent discrimination for myocardial infarction (area under curve 0.953, 95% confidence interval 0.947-0.958), performed well across subgroups, and identified more patients at presentation as low59 probability as having myocardial infarction than fixed cardiac troponin thresholds (61% versus 27%) with a similar negative predictive value, and fewer as high-probability for having myocardial infarction (10% versus 16%) with a greater positive predictive value. Patients identified as having a low-probability of myocardial infarction had a lower rate of cardiac death than those with intermediate- or high-probability 30-days (0.1% versus 0.5% and 1.8%) and one year (0.3% versus 2.8% and 4.2%; P<0.001 for both) from patient presentation. CoDE-ACS used as a clinical decision support system has the potential to reduce hospital admissions and have major benefits for patients and healthcare providers.

Citation

Doudesis, D., Lee, K. K., Boeddinghaus, J., Bularga, A., Ferry, A. V., Tuck, C., Lowry, M. T., Lopez-Ayala, P., Nestelberger, T., Koechlin, L., Bernabeu, M. O., Neubeck, L., Anand, A., Schulz, K., Apple, F. S., Parsonage, W., Greenslade, J. H., Cullen, L., Pickering, J. W., Than, M. P., …CoDE-ACS Investigators. (2023). Machine learning for diagnosis of myocardial infarction using cardiac troponin concentrations. Nature Medicine, 29(5), 1201-1210. https://doi.org/10.1038/s41591-023-02325-4

• Journal Article Type: Article
• Acceptance Date: Mar 29, 2023
• Online Publication Date: May 11, 2023
• Publication Date: 2023
• Deposit Date: Mar 30, 2023
• Publicly Available Date: May 11, 2023
• Journal: Nature Medicine
• Print ISSN: 1078-8956
• Electronic ISSN: 1546-170X
• Publisher: Nature Research
• Peer Reviewed: Peer Reviewed
• Volume: 29
• Issue: 5
• Pages: 1201-1210
• DOI: https://doi.org/10.1038/s41591-023-02325-4