A Lightweight LCDECG Model for Cardiovascular Diagnostics Using ECG Features

Title

A Lightweight LCDECG Model for Cardiovascular Diagnostics Using ECG Features

Creator

Nitin Joshua, M.; Kumar, Pawan; George, Jossy

Description

Cardiovascular disease (CVD) is among the leading causes of death around the world, requiring accurate and reliable diagnostics, and early detection. This project aims at the development of an efficient and accurate, lightweight model to classify heart rhythms based on an ECG.. In this paper, we propose the LCDECG (Lightweight Cardiac Diagnostic ECG) model, which integrates deep morphological feature extraction from ECG with clinically relevant handcrafted features. With MobileNetV2 used as a feature extractor and statistical descriptors of ECG signals, both the pathways are combined at the feature level for multi-class classification of cardiac conditions. Experiments conducted on the Dataset demonstrate better classification performance with 97.8% accuracy, 96.4% precision, 97.1% recall, and 96.7% F1-score over traditional neural networks alone or only statistical methods. The model is able to achieve the desired results as it only utilizes 2.43M parameters in its architecture, and therefore is amenable to real-time deployment in resource-scarce environments. Its use is advantageous for facilitating timely and early detection, which is necessary to improve patient survival and reduce healthcare costs through preventative treatment. Current ECG readings are based on manual assessment by trained cardiologists, which can be time-consuming and potentially subjective, depending on several professionals in the medical field evaluating the tracing. Due to the increased incidence of cardiovascular disease globally, and the limited number of professionals, particularly in developing countries, there is even greater need for automated convenient and trustworthy ECG tracing for diagnostic support. 2025 IEEE.

Source

2025 IEEE 6th Global Conference for Advancement in Technology, GCAT 2025;

Date

01-01-2025

Publisher

Institute of Electrical and Electronics Engineers Inc.

Subject

Cardiovascular Disease (CVD); Deep learning; Electrocardiogram (ECG); Feature fusion; Lightweight models; MobileNetV2; Real-time diagnosis; Statistical descriptors

Coverage

Nitin Joshua M., Christ (Deemed to be University), India; Kumar P., Christ (Deemed to be University), India; George J., Christ (Deemed to be University), India

Rights

Restricted Access; Hardcopy may be available in the library

Relation

ISBN: 979-833151458-7;

Format

online

Language

English

Type

Conference paper

Identifier

https://doi.org/10.1109/GCAT66372.2025.11368610

https://www.scopus.com/pages/publications/105034258168?origin=resultslist