A training pipeline of an arrhythmia classifier for atrial fibrillation detection using Photoplethysmography signal
Photoplethysmography (PPG) signal is potentially suitable in atrial fibrillation (AF) detection for its convenience in use and similarity in physiological origin to electrocardiogram (ECG). There are a few preceding studies that have shown the possibility of using the peak-to-peak interval of the PP...
Main Authors: | , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Frontiers Media S.A.
2023-01-01
|
Series: | Frontiers in Physiology |
Subjects: | |
Online Access: | https://www.frontiersin.org/articles/10.3389/fphys.2023.1084837/full |
_version_ | 1797948260034281472 |
---|---|
author | Sota Kudo Zheng Chen Xue Zhou Leighton T. Izu Ye Chen-Izu Xin Zhu Toshiyo Tamura Shigehiko Kanaya Ming Huang |
author_facet | Sota Kudo Zheng Chen Xue Zhou Leighton T. Izu Ye Chen-Izu Xin Zhu Toshiyo Tamura Shigehiko Kanaya Ming Huang |
author_sort | Sota Kudo |
collection | DOAJ |
description | Photoplethysmography (PPG) signal is potentially suitable in atrial fibrillation (AF) detection for its convenience in use and similarity in physiological origin to electrocardiogram (ECG). There are a few preceding studies that have shown the possibility of using the peak-to-peak interval of the PPG signal (PPIp) in AF detection. However, as a generalized model, the accuracy of an AF detector should be pursued on the one hand; on the other hand, its generalizability should be paid attention to in view of the individual differences in PPG manifestation of even the same arrhythmia and the existence of sub-types. Moreover, a binary classifier for atrial fibrillation and normal sinus rhythm is not convincing enough for the similarity between AF and ectopic beats. In this study, we project the atrial fibrillation detection as a multiple-class classification and try to propose a training pipeline that is advantageous both to the accuracy and generalizability of the classifier by designing and determining the configurable options of the pipeline, in terms of input format, deep learning model (with hyperparameter optimization), and scheme of transfer learning. With a rigorous comparison of the possible combinations of the configurable components in the pipeline, we confirmed that first-order difference of heartbeat sequence as the input format, a 2-layer CNN–1-layer Transformer hybridR model as the learning model and the whole model fine-tuning as the implementing scheme of transfer learning is the best combination for the pipeline (F1 value: 0.80, overall accuracy: 0.87)R. |
first_indexed | 2024-04-10T21:40:35Z |
format | Article |
id | doaj.art-9d7df1295d714287868c3d2c5150e593 |
institution | Directory Open Access Journal |
issn | 1664-042X |
language | English |
last_indexed | 2024-04-10T21:40:35Z |
publishDate | 2023-01-01 |
publisher | Frontiers Media S.A. |
record_format | Article |
series | Frontiers in Physiology |
spelling | doaj.art-9d7df1295d714287868c3d2c5150e5932023-01-19T05:42:54ZengFrontiers Media S.A.Frontiers in Physiology1664-042X2023-01-011410.3389/fphys.2023.10848371084837A training pipeline of an arrhythmia classifier for atrial fibrillation detection using Photoplethysmography signalSota Kudo0Zheng Chen1Xue Zhou2Leighton T. Izu3Ye Chen-Izu4Xin Zhu5Toshiyo Tamura6Shigehiko Kanaya7Ming Huang8Computational Systems Biology Lab, Graduate School of Science and Technology, Nara Institute of Science and Technology, Ikoma, JapanISIR, Osaka University, Osaka, JapanComputational Systems Biology Lab, Graduate School of Science and Technology, Nara Institute of Science and Technology, Ikoma, JapanDepartment of Pharmacology, University of California, Davis, Davis, CA, United StatesDepartment of Biomedical Engineering, University of California, Davis, Davis, CA, United StatesBiomedical Information Engineering Lab, The University of Aizu, Aizu-Wakamatsu, JapanFuture Robotics Organization, Waseda University, Tokyo, JapanComputational Systems Biology Lab, Graduate School of Science and Technology, Nara Institute of Science and Technology, Ikoma, JapanComputational Systems Biology Lab, Graduate School of Science and Technology, Nara Institute of Science and Technology, Ikoma, JapanPhotoplethysmography (PPG) signal is potentially suitable in atrial fibrillation (AF) detection for its convenience in use and similarity in physiological origin to electrocardiogram (ECG). There are a few preceding studies that have shown the possibility of using the peak-to-peak interval of the PPG signal (PPIp) in AF detection. However, as a generalized model, the accuracy of an AF detector should be pursued on the one hand; on the other hand, its generalizability should be paid attention to in view of the individual differences in PPG manifestation of even the same arrhythmia and the existence of sub-types. Moreover, a binary classifier for atrial fibrillation and normal sinus rhythm is not convincing enough for the similarity between AF and ectopic beats. In this study, we project the atrial fibrillation detection as a multiple-class classification and try to propose a training pipeline that is advantageous both to the accuracy and generalizability of the classifier by designing and determining the configurable options of the pipeline, in terms of input format, deep learning model (with hyperparameter optimization), and scheme of transfer learning. With a rigorous comparison of the possible combinations of the configurable components in the pipeline, we confirmed that first-order difference of heartbeat sequence as the input format, a 2-layer CNN–1-layer Transformer hybridR model as the learning model and the whole model fine-tuning as the implementing scheme of transfer learning is the best combination for the pipeline (F1 value: 0.80, overall accuracy: 0.87)R.https://www.frontiersin.org/articles/10.3389/fphys.2023.1084837/fullatrial fibrillationectopic beatsnormal sinus rhythmdeep learningartificial neural networkmodel generalizability |
spellingShingle | Sota Kudo Zheng Chen Xue Zhou Leighton T. Izu Ye Chen-Izu Xin Zhu Toshiyo Tamura Shigehiko Kanaya Ming Huang A training pipeline of an arrhythmia classifier for atrial fibrillation detection using Photoplethysmography signal Frontiers in Physiology atrial fibrillation ectopic beats normal sinus rhythm deep learning artificial neural network model generalizability |
title | A training pipeline of an arrhythmia classifier for atrial fibrillation detection using Photoplethysmography signal |
title_full | A training pipeline of an arrhythmia classifier for atrial fibrillation detection using Photoplethysmography signal |
title_fullStr | A training pipeline of an arrhythmia classifier for atrial fibrillation detection using Photoplethysmography signal |
title_full_unstemmed | A training pipeline of an arrhythmia classifier for atrial fibrillation detection using Photoplethysmography signal |
title_short | A training pipeline of an arrhythmia classifier for atrial fibrillation detection using Photoplethysmography signal |
title_sort | training pipeline of an arrhythmia classifier for atrial fibrillation detection using photoplethysmography signal |
topic | atrial fibrillation ectopic beats normal sinus rhythm deep learning artificial neural network model generalizability |
url | https://www.frontiersin.org/articles/10.3389/fphys.2023.1084837/full |
work_keys_str_mv | AT sotakudo atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT zhengchen atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT xuezhou atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT leightontizu atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT yechenizu atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT xinzhu atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT toshiyotamura atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT shigehikokanaya atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT minghuang atrainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT sotakudo trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT zhengchen trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT xuezhou trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT leightontizu trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT yechenizu trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT xinzhu trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT toshiyotamura trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT shigehikokanaya trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal AT minghuang trainingpipelineofanarrhythmiaclassifierforatrialfibrillationdetectionusingphotoplethysmographysignal |