AI-Based ECG Analysis for Early Myocardial Infarction Detection

Krima Ashokkumar Patel

International Journal of Scientific Research in Computer Science, Engineering and Information Technology, 2024, vol. 10, issue 6, 2703-2709

Abstract: The 12-lead electrocardiogram is the most widely used first line test for heart disease, but expert review is a bottleneck in busy clinics. This paper compares two deep models for detection of myocardial infarction against confirmed normal recordings on the PTB-XL corpus. The first is a 1D Residual Network with eight residual blocks over the raw signal. The second is a convolutional front end followed by a bidirectional LSTM with attention pooling. Both models are trained end to end on the standard PTB-XL split with light augmentation and class weighted cross entropy. On the held out fold the ResNet reaches 0.9630 accuracy and AUROC 0.981, and the recurrent model reaches 0.9520 accuracy and AUROC 0.981. Both improve over a strong handcrafted feature baseline. The models fit on a single RTX 6000 Ada GPU and train in under two minutes each.

Keywords: Electrocardiogram; Deep Learning; Residual Network; BiLSTM; Attention; Myocardial Infarction (search for similar items in EconPapers)
Date: 2024
Note: Article URL: https://ijsrcseit.com/home/article/view/CSEIT2410790
References: Add references at CitEc
Citations:

Downloads: (external link)
https://ijsrcseit.com/home/article/view/CSEIT2410790 Article URL (text/html)
https://ijsrcseit.com/home/article/download/CSEIT2410790/CSEIT2410790 Full text (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:jbh:ijsrcs:v10:y2024:i6:id:1997

DOI: 10.32628/CSEIT2410790

Access Statistics for this article

More articles in International Journal of Scientific Research in Computer Science, Engineering and Information Technology from International Journal of Scientific Research in Computer Science, Engineering and Information Technology
Bibliographic data for series maintained by Pankaj Sharma (USA) ().