Multi-task deep learning for cardiac rhythm detection in wearable devices

Abstract Wearable devices enable theoretically continuous, longitudinal monitoring of physiological measurements such as step count, energy expenditure, and heart rate. Although the classification of abnormal cardiac rhythms such as atrial fibrillation from wearable devices has great potential, comm...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Jessica Torres-Soto, Euan A. Ashley
Formato:	article
Lenguaje:	EN
Publicado:	Nature Portfolio 2020
Materias:	Computer applications to medicine. Medical informatics R858-859.7
Acceso en línea:	https://doaj.org/article/5a718835d2b145be865ed6e1fbd315a6
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:5a718835d2b145be865ed6e1fbd315a6
record_format	dspace
spelling	oai:doaj.org-article:5a718835d2b145be865ed6e1fbd315a62021-12-02T19:13:53ZMulti-task deep learning for cardiac rhythm detection in wearable devices10.1038/s41746-020-00320-42398-6352https://doaj.org/article/5a718835d2b145be865ed6e1fbd315a62020-09-01T00:00:00Zhttps://doi.org/10.1038/s41746-020-00320-4https://doaj.org/toc/2398-6352Abstract Wearable devices enable theoretically continuous, longitudinal monitoring of physiological measurements such as step count, energy expenditure, and heart rate. Although the classification of abnormal cardiac rhythms such as atrial fibrillation from wearable devices has great potential, commercial algorithms remain proprietary and tend to focus on heart rate variability derived from green spectrum LED sensors placed on the wrist, where noise remains an unsolved problem. Here we develop DeepBeat, a multitask deep learning method to jointly assess signal quality and arrhythmia event detection in wearable photoplethysmography devices for real-time detection of atrial fibrillation. The model is trained on approximately one million simulated unlabeled physiological signals and fine-tuned on a curated dataset of over 500 K labeled signals from over 100 individuals from 3 different wearable devices. We demonstrate that, in comparison with a single-task model, our architecture using unsupervised transfer learning through convolutional denoising autoencoders dramatically improves the performance of atrial fibrillation detection from a F1 score of 0.54 to 0.96. We also include in our evaluation a prospectively derived replication cohort of ambulatory participants where the algorithm performed with high sensitivity (0.98), specificity (0.99), and F1 score (0.93). We show that two-stage training can help address the unbalanced data problem common to biomedical applications, where large-scale well-annotated datasets are hard to generate due to the expense of manual annotation, data acquisition, and participant privacy.Jessica Torres-SotoEuan A. AshleyNature PortfolioarticleComputer applications to medicine. Medical informaticsR858-859.7ENnpj Digital Medicine, Vol 3, Iss 1, Pp 1-8 (2020)
institution	DOAJ
collection	DOAJ
language	EN
topic	Computer applications to medicine. Medical informatics R858-859.7
spellingShingle	Computer applications to medicine. Medical informatics R858-859.7 Jessica Torres-Soto Euan A. Ashley Multi-task deep learning for cardiac rhythm detection in wearable devices
description	Abstract Wearable devices enable theoretically continuous, longitudinal monitoring of physiological measurements such as step count, energy expenditure, and heart rate. Although the classification of abnormal cardiac rhythms such as atrial fibrillation from wearable devices has great potential, commercial algorithms remain proprietary and tend to focus on heart rate variability derived from green spectrum LED sensors placed on the wrist, where noise remains an unsolved problem. Here we develop DeepBeat, a multitask deep learning method to jointly assess signal quality and arrhythmia event detection in wearable photoplethysmography devices for real-time detection of atrial fibrillation. The model is trained on approximately one million simulated unlabeled physiological signals and fine-tuned on a curated dataset of over 500 K labeled signals from over 100 individuals from 3 different wearable devices. We demonstrate that, in comparison with a single-task model, our architecture using unsupervised transfer learning through convolutional denoising autoencoders dramatically improves the performance of atrial fibrillation detection from a F1 score of 0.54 to 0.96. We also include in our evaluation a prospectively derived replication cohort of ambulatory participants where the algorithm performed with high sensitivity (0.98), specificity (0.99), and F1 score (0.93). We show that two-stage training can help address the unbalanced data problem common to biomedical applications, where large-scale well-annotated datasets are hard to generate due to the expense of manual annotation, data acquisition, and participant privacy.
format	article
author	Jessica Torres-Soto Euan A. Ashley
author_facet	Jessica Torres-Soto Euan A. Ashley
author_sort	Jessica Torres-Soto
title	Multi-task deep learning for cardiac rhythm detection in wearable devices
title_short	Multi-task deep learning for cardiac rhythm detection in wearable devices
title_full	Multi-task deep learning for cardiac rhythm detection in wearable devices
title_fullStr	Multi-task deep learning for cardiac rhythm detection in wearable devices
title_full_unstemmed	Multi-task deep learning for cardiac rhythm detection in wearable devices
title_sort	multi-task deep learning for cardiac rhythm detection in wearable devices
publisher	Nature Portfolio
publishDate	2020
url	https://doaj.org/article/5a718835d2b145be865ed6e1fbd315a6
work_keys_str_mv	AT jessicatorressoto multitaskdeeplearningforcardiacrhythmdetectioninwearabledevices AT euanaashley multitaskdeeplearningforcardiacrhythmdetectioninwearabledevices
_version_	1718377019270496256

Multi-task deep learning for cardiac rhythm detection in wearable devices

Ejemplares similares