IEEE J Biomed Health Inform. 2024 Nov;28(11):6525-6537. doi: 10.1109/JBHI.2024.3456028. Epub 2024 Nov 6.
This systematic review aims to summarize the consumer wearable devices used for collecting ECG signals, explore the models or algorithms employed in diagnosing and preventing heart-related diseases through ECG analysis, and discuss the challenges and future work related to adopting health monitoring using consumer wearable devices. Following the PRISMA method, we identified and reviewed 102 relevant papers from PubMed, IEEE, and Web of Science databases, covering the period from May 2013 to May 2023. This review comprehensively summarizes consumer wearable devices with ECG functions, available ECG datasets, and various algorithms for detecting cardiac diseases and monitoring long-term health. It also discusses the integration challenges and future directions in cardiac health monitoring. The results highlight a preference for deep learning algorithms, such as Convolutional Neural Networks (CNNs) and their variations, in analyzing ECG data due to the ability to automate feature extraction and reduce memory requirements. The review also discusses potential limitations of the current literature, including lack of reasoning and comparison of algorithms and limited data generalizability. By analyzing the current literature, this review provides an overview of state-of-the-art technologies, identifies key findings, and suggests potential avenues for future research and implementation.
本系统评价旨在总结用于采集 ECG 信号的消费者可穿戴设备，通过 ECG 分析探索用于诊断和预防与心脏相关疾病的模型或算法，并讨论采用消费者可穿戴设备进行健康监测的相关挑战和未来工作。我们按照 PRISMA 方法，从 PubMed、IEEE 和 Web of Science 数据库中确定并回顾了 102 篇相关文献，涵盖了 2013 年 5 月至 2023 年 5 月的时间段。本评价全面总结了具有 ECG 功能的消费者可穿戴设备、可用的 ECG 数据集以及用于检测心脏疾病和监测长期健康的各种算法。它还讨论了心脏健康监测中的集成挑战和未来方向。结果强调了深度学习算法（如卷积神经网络 (CNN) 及其变体）在分析 ECG 数据方面的偏好，因为这些算法能够自动化特征提取并降低内存需求。评价还讨论了当前文献的潜在局限性，包括缺乏算法的推理和比较以及数据通用性有限。通过分析当前文献，本评价提供了最先进技术的概述，确定了关键发现，并提出了未来研究和实施的潜在途径。
