Smart ECG Analysis and Heart Condition Detection Using AI

Abstract

Cardiovascular diseases (CVDs) remain the leading cause of mortality worldwide, accounting for an estimated 17.9 million deaths annually. Electrocardiography (ECG) is a non-invasive and widely utilized diagnostic tool for detecting cardiac abnormalities. However, manual interpretation of ECG signals is time-consuming and prone to error. In this study, we propose a deep learning–based framework for automated ECG analysis and heart condition detection.Utilizing the large-scale PTB-XL 12-lead ECG dataset, we apply essential preprocessing steps including bandpass filtering and baseline drift correction to enhance signal quality. Our architecture integrates a one dimensional convolutional neural network (1D-CNN) for morphological feature extraction with a bidirectional long short-term memory (BiLSTM) network to capture temporal dependencies. Additionally, patient demographic information (age and sex) is incorporated through a parallel input branch to improve classification robustness.The model is trained using categorical cross-entropy loss across five diagnostic classes: normal, myocardial infarction, ST/T abnormalities, conduction disturbances, and hypertrophy. Stratified data splitting into training, validation, and test sets ensures balanced evaluation. The final model achieved a test accuracy of approximately 77.9%, with class-wise F1-scores ranging from 0.70 to 0.80, indicating reliable detection across diagnostic categories. A confusion matrix highlights misclassifications, particularly among classes with overlapping waveform characteristics.This AI-enhanced ECG interpretation system offers a consistent, scalable solution for preliminary cardiac screening, with strong potential for deployment in telehealth and resource-constrained environments to support early diagnosis and intervention.