Anomaly Detection in Multi-Wavelength Photoplethysmography Using Lightweight Machine Learning Algorithms

Vlad-Eusebiu Baciu; Joan Lambert Cause; Angel Sole Morillo; Juan Carlos Garcia-Naranjo; Johan Stiens; Bruno da Silva

doi:10.3390/s23156947

Anomaly Detection in Multi-Wavelength Photoplethysmography Using Lightweight Machine Learning Algorithms

Vlad-Eusebiu Baciu, Joan Lambert Cause, Angel Sole Morillo, Juan Carlos Garcia-Naranjo, Johan Stiens, Bruno da Silva

Research output: Contribution to journal › Article › peer-review

46 Downloads (Pure)

Abstract

Over the past few years, there has been increased interest in photoplethysmography (PPG) technology, which has revealed that, in addition to heart rate and oxygen saturation, the pulse shape of the PPG signal contains much more valuable information. Lately, the wearable market has shifted towards a multi-wavelength and multichannel approach to increase signal robustness and facilitate the extraction of other intrinsic information from the signal. This transition presents several challenges related to complexity, accuracy, and reliability of algorithms. To address these challenges, anomaly detection stages can be employed to increase the accuracy and reliability of estimated parameters. Powerful algorithms, such as lightweight machine learning (ML) algorithms, can be used for anomaly detection in multi-wavelength PPG (MW-PPG). The main contributions of this paper are (a) proposing a set of features with high information gain for anomaly detection in MW-PPG signals in the classification context, (b) assessing the impact of window size and evaluating various lightweight ML models to achieve highly accurate anomaly detection, and (c) examining the effectiveness of MW-PPG signals in detecting artifacts.

Original language	English
Article number	6947
Pages (from-to)	1-23
Number of pages <span style="color:red"p> <font size="1.5"> ✽ </span> </font>	23
Journal	Sensors
Volume	23
Issue number	15
DOIs	https://doi.org/10.3390/s23156947
Publication status	Published - 4 Aug 2023

Bibliographical note

Funding Information:
This work has been possible thanks to the Post-doc VUB Global Minds 2021 call approved as VLIR396. This work has been partially supported by the Vrije Universiteit Brussel (VUB) through the SRP-Projects M3D2 and LSDS, the ETRO-IOF Project under Grant IOF3016; partially by the FWO-Flanders FWOSB106 PhD grant. Finally, this work has been also partially supported by the Belgian Development Cooperation through VLIR-UOS (Flemish Interuniversity Council-University Cooperation for Development) in the context of the Institutional University Cooperation program (IUC 2019 Phase 2 UO) with the Universidad de Oriente (Cuba).

Publisher Copyright:
© 2023 by the authors.

Keywords

photoplethysmography
PPG
artifact
anomaly detection
multi-wavelength PPG
time series
machine learning
neural networks
supervised learning
unsupervised learning

Access to Document

10.3390/s23156947Licence: CC BY

99056818Final published version, 1.26 MBLicence: CC BY

Cite this

@article{7082997c9b4349a0b6edca34bb0c218a,

title = "Anomaly Detection in Multi-Wavelength Photoplethysmography Using Lightweight Machine Learning Algorithms",

abstract = "Over the past few years, there has been increased interest in photoplethysmography (PPG) technology, which has revealed that, in addition to heart rate and oxygen saturation, the pulse shape of the PPG signal contains much more valuable information. Lately, the wearable market has shifted towards a multi-wavelength and multichannel approach to increase signal robustness and facilitate the extraction of other intrinsic information from the signal. This transition presents several challenges related to complexity, accuracy, and reliability of algorithms. To address these challenges, anomaly detection stages can be employed to increase the accuracy and reliability of estimated parameters. Powerful algorithms, such as lightweight machine learning (ML) algorithms, can be used for anomaly detection in multi-wavelength PPG (MW-PPG). The main contributions of this paper are (a) proposing a set of features with high information gain for anomaly detection in MW-PPG signals in the classification context, (b) assessing the impact of window size and evaluating various lightweight ML models to achieve highly accurate anomaly detection, and (c) examining the effectiveness of MW-PPG signals in detecting artifacts.",

keywords = "photoplethysmography, PPG, artifact, anomaly detection, multi-wavelength PPG, time series, machine learning, neural networks, supervised learning, unsupervised learning",

author = "Vlad-Eusebiu Baciu and {Lambert Cause}, Joan and {Sole Morillo}, Angel and Garcia-Naranjo, {Juan Carlos} and Johan Stiens and {da Silva}, Bruno",

note = "Funding Information: This work has been possible thanks to the Post-doc VUB Global Minds 2021 call approved as VLIR396. This work has been partially supported by the Vrije Universiteit Brussel (VUB) through the SRP-Projects M3D2 and LSDS, the ETRO-IOF Project under Grant IOF3016; partially by the FWO-Flanders FWOSB106 PhD grant. Finally, this work has been also partially supported by the Belgian Development Cooperation through VLIR-UOS (Flemish Interuniversity Council-University Cooperation for Development) in the context of the Institutional University Cooperation program (IUC 2019 Phase 2 UO) with the Universidad de Oriente (Cuba). Publisher Copyright: {\textcopyright} 2023 by the authors.",

year = "2023",

month = aug,

day = "4",

doi = "10.3390/s23156947",

language = "English",

volume = "23",

pages = "1--23",

journal = "Sensors",

issn = "1424-8220",

publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",

number = "15",

}

TY - JOUR

T1 - Anomaly Detection in Multi-Wavelength Photoplethysmography Using Lightweight Machine Learning Algorithms

AU - Baciu, Vlad-Eusebiu

AU - Lambert Cause, Joan

AU - Sole Morillo, Angel

AU - Garcia-Naranjo, Juan Carlos

AU - Stiens, Johan

AU - da Silva, Bruno

N1 - Funding Information: This work has been possible thanks to the Post-doc VUB Global Minds 2021 call approved as VLIR396. This work has been partially supported by the Vrije Universiteit Brussel (VUB) through the SRP-Projects M3D2 and LSDS, the ETRO-IOF Project under Grant IOF3016; partially by the FWO-Flanders FWOSB106 PhD grant. Finally, this work has been also partially supported by the Belgian Development Cooperation through VLIR-UOS (Flemish Interuniversity Council-University Cooperation for Development) in the context of the Institutional University Cooperation program (IUC 2019 Phase 2 UO) with the Universidad de Oriente (Cuba). Publisher Copyright: © 2023 by the authors.

PY - 2023/8/4

Y1 - 2023/8/4

N2 - Over the past few years, there has been increased interest in photoplethysmography (PPG) technology, which has revealed that, in addition to heart rate and oxygen saturation, the pulse shape of the PPG signal contains much more valuable information. Lately, the wearable market has shifted towards a multi-wavelength and multichannel approach to increase signal robustness and facilitate the extraction of other intrinsic information from the signal. This transition presents several challenges related to complexity, accuracy, and reliability of algorithms. To address these challenges, anomaly detection stages can be employed to increase the accuracy and reliability of estimated parameters. Powerful algorithms, such as lightweight machine learning (ML) algorithms, can be used for anomaly detection in multi-wavelength PPG (MW-PPG). The main contributions of this paper are (a) proposing a set of features with high information gain for anomaly detection in MW-PPG signals in the classification context, (b) assessing the impact of window size and evaluating various lightweight ML models to achieve highly accurate anomaly detection, and (c) examining the effectiveness of MW-PPG signals in detecting artifacts.

AB - Over the past few years, there has been increased interest in photoplethysmography (PPG) technology, which has revealed that, in addition to heart rate and oxygen saturation, the pulse shape of the PPG signal contains much more valuable information. Lately, the wearable market has shifted towards a multi-wavelength and multichannel approach to increase signal robustness and facilitate the extraction of other intrinsic information from the signal. This transition presents several challenges related to complexity, accuracy, and reliability of algorithms. To address these challenges, anomaly detection stages can be employed to increase the accuracy and reliability of estimated parameters. Powerful algorithms, such as lightweight machine learning (ML) algorithms, can be used for anomaly detection in multi-wavelength PPG (MW-PPG). The main contributions of this paper are (a) proposing a set of features with high information gain for anomaly detection in MW-PPG signals in the classification context, (b) assessing the impact of window size and evaluating various lightweight ML models to achieve highly accurate anomaly detection, and (c) examining the effectiveness of MW-PPG signals in detecting artifacts.

KW - photoplethysmography

KW - PPG

KW - artifact

KW - anomaly detection

KW - multi-wavelength PPG

KW - time series

KW - machine learning

KW - neural networks

KW - supervised learning

KW - unsupervised learning

U2 - 10.3390/s23156947

DO - 10.3390/s23156947

M3 - Article

VL - 23

SP - 1

EP - 23

JO - Sensors

JF - Sensors

SN - 1424-8220

IS - 15

M1 - 6947

ER -

Anomaly Detection in Multi-Wavelength Photoplethysmography Using Lightweight Machine Learning Algorithms

Abstract

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this