Automatic electrocardiographic algorithm for assessing severity of ischemia in ST-segment elevation myocardial infarction

作者:Fakhri Yama*; Melgaard Jacob; Andersson Hedvig Bille; Schoos Mikkel Malby; Birnbaum Yochai; Graff Claus; Sejersten Maria; Kastrup Jens; Clemmensen Peter
来源:International Journal of Cardiology, 2018, 268: 18-22.
DOI:10.1016/j.ijcard.2018.04.057

摘要

Background: Terminal QRS distortion on the electrocardiogram(ECG) is a sign of severe ischemia in patients with STEMI and can be quantified by the Sclarovsky-Birnbaum Severity of Ischemia. Due to score complexity, it has not been applied in clinical practice. Automatic scoring of digitally recorded ECGs could facilitate clinical application. We aimed to develop an automatic algorithm for the severity of ischemia.
Methods: Development set: 50 STEMI ECGs were manually (Manual-score) and automatically (Auto-score) scored by our designed algorithm. The agreement between Manual-and Auto-score was assessed by kappa statistics. Test set: ECGs from 199 STEMI patients were assigned a severity grade (severe or non-severe ischemia) by the Auto-score. Infarct size estimated by median peak Troponin T (TnT) and Creatinine Kinase Myocardial Band (CKMB) was tested between the groups.
Results: The agreement between Manual- and Auto-score was 0.83 ((95% CI 0.55-1.00), p < 0.0001), sensitivity 75% and specificity 100%, PPV 100% and NPV 94.6%. In the test set 152 (76%) patients were male, mean age 61 +/- 12 years. The Auto-score designated severe ischemia in 42 (21%) and non-severe ischemia in 157 (79%) patients. Patients with ECG signs of severe vs. non-severe ischemia had significantly higher levels of biomarkers of infarct size. In multiple linear regression, ECG sign of severe ischemia was an independent predictor for higher TnT and CKMB levels.
Conclusion: The automatic ECG algorithm for severity of ischemia in STEMI performs adequately for clinical use. Severe ischemia obtained by the Auto-score was associated with biomarker estimated larger infarct size.

  • 出版日期2018-10-1