For this project, I built a deep architecture consisting one-dimensional Convolutional Neural Network (1D-CNN) and bidirectional Long-Short Term Memory Network (biLSTM) layers for automated detection and localization of Myocardial Infarction (MI) ailment. As to the data-preprocessing part, I considered extracting thirty-six morphological and time-domain features from 12-lead electrocardiogram (ECG) beats (amplitude of T-wave, amplitude of Q-wave, and ST-segment deviation for all leads), acquired from the PTBXL 2021 dataset.
"THE FINAL ACCURACY OBTAINED = 95.33% to 95.72% (depending on the type of data-preprocessing)"
Refer to the Final Research Report for profound insights into my research work and code-execution.