Deep CNN-based detection of cardiac rhythm disorders using PPG signals from wearable devices

Miray Gunay Bulut, Sencer Unal, Mohamed Hammad, Paweł Pławiak

Abstract

Cardiac rhythm disorders can manifest in various ways, such as the heart rate being too fast (tachycardia) or too slow (bradycardia), irregular heartbeats (like atrial fibrillation-AF, ventricular fibrillation-VF), or the initiation of heartbeats in different areas from the norm (extrasystole).

Introduction

Cardiac arrhythmias are abnormal heart rhythms that can lead to serious health problems such as strokes, heart failure, and even death. Therefore, early detection of arrhythmias is important for effective treatment and management [1].

Materials and methods

In this article, arrhythmia detection has been achieved using a deep learning model on PPG signals obtained from wearable devices. For this purpose, a dataset consisting of signals from 37 subjects, including both ECG and PPG signals, was utilized.

Results

The methods tested in this study were developed using the Python language in the Google Colab environment. The ECG and PPG datasets, along with the classes they contain, have an equal number of data points and were split into training (70%), validation (15%), and testing (15%) sets, as shown in Table 2.

Discussion

In this study, a PPG dataset prepared by the University of Massachusetts Medical Center (UMMC) was utilized. Table 4 provides a summary of a comparison of studies developed for the detection of cardiovascular diseases from signals obtained from wearable devices with machine learning methods.

Conclusions

Taking heart signals from patients using an electrocardiogram requires patients to visit a healthcare facility. Wearable devices can overcome this challenge by continuously recording and monitoring signals from the heart using Photoplethysmography technology, allowing us to gain insights into heart health wherever we are.

Citation: Bulut MG, Unal S, Hammad M, Pławiak P (2025) Deep CNN-based detection of cardiac rhythm disorders using PPG signals from wearable devices. PLoS ONE 20(2): e0314154. https://doi.org/10.1371/journal.pone.0314154

Editor: Agnese Sbrollini, Polytechnic University of Marche: Universita Politecnica delle Marche, ITALY

Received: January 19, 2024; Accepted: November 5, 2024; Published: February 12, 2025

Copyright: © 2025 Bulut et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: This study utilized third party data from the UMMC Simband Dataset for Pulsewatch (https://www.synapse.org/pulsewatch). The authors do not have permission to publicly share minimal data for this study from UMMC Simband Dataset. Data are available upon request from UMMC representative, Dong Han (Department of Biomedical Engineering, University of Connecticut, Storrs, CT 06269, USA), via email (dong.han@uconn.edu) for researchers. Researchers are required to register a Synapse account (https://accounts.synapse.org/register1), send a request email to Dong Han with your Synapse ID, real name, and organizational affiliation to access these data. The authors confirm that others would be able to access or request these data in the same manner as themselves previously described. The authors also confirm that they did not have any special access or request privileges that others would not have.

Funding: The author(s) received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

 

Source: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0314154#abstract0