TY - JOUR

T1 - Fitbeat

T2 - COVID-19 estimation based on wristband heart rate using a contrastive convolutional auto-encoder

AU - Liu, Shuo

AU - Han, Jing

AU - Puyal, Estela Laporta

AU - Kontaxis, Spyridon

AU - Sun, Shaoxiong

AU - Locatelli, Patrick

AU - Dineley, Judith

AU - Pokorny, Florian B.

AU - Dalla Costa, Gloria

AU - Leocani, Letizia

AU - Guerrero, Ana Isabel

AU - Nos, Carlos

AU - Zabalza, Ana

AU - Sorensen, Per Soelberg

AU - Buron, Mathias

AU - Magyari, Melinda

AU - Ranjan, Yatharth

AU - Rashid, Zulqarnain

AU - Conde, Pauline

AU - Stewart, Callum

AU - Folarin, Amos A.

AU - Dobson, Richard J. B.

AU - Bailon, Raquel

AU - Vairavan, Srinivasan

AU - Cummins, Nicholas

AU - Narayan, Vaibhav A.

AU - Hotopf, Matthew

AU - Comi, Giancarlo

AU - Schuller, Bjoern

AU - RADAR-CNS Consortium

PY - 2022/3

Y1 - 2022/3

N2 - This study proposes a contrastive convolutional auto-encoder (contrastive CAE), a combined architecture of an auto-encoder and contrastive loss, to identify individuals with suspected COVID-19 infection using heart-rate data from participants with multiple sclerosis (MS) in the ongoing RADAR-CNS mHealth research project. Heart-rate data was remotely collected using a Fitbit wristband. COVID-19 infection was either confirmed through a positive swab test, or inferred through a self-reported set of recognised symptoms of the virus. The contrastive CAE outperforms a conventional convolutional neural network (CNN), a long short-term memory (LSTM) model, and a convolutional auto-encoder without contrastive loss (CAE). On a test set of 19 participants with MS with reported symptoms of COVID-19, each one paired with a participant with MS with no COVID-19 symptoms, the contrastive CAE achieves an unweighted average recall of 95 . 3% , a sensitivity of 100% and a specificity of 90 . 6% , an area under the receiver operating characteristic curve (AUC-ROC) of 0.944, indicating a maximum successful detection of symptoms in the given heart rate measurement period, whilst at the same time keeping a low false alarm rate. (c) 2021 Elsevier Ltd. All rights reserved.

AB - This study proposes a contrastive convolutional auto-encoder (contrastive CAE), a combined architecture of an auto-encoder and contrastive loss, to identify individuals with suspected COVID-19 infection using heart-rate data from participants with multiple sclerosis (MS) in the ongoing RADAR-CNS mHealth research project. Heart-rate data was remotely collected using a Fitbit wristband. COVID-19 infection was either confirmed through a positive swab test, or inferred through a self-reported set of recognised symptoms of the virus. The contrastive CAE outperforms a conventional convolutional neural network (CNN), a long short-term memory (LSTM) model, and a convolutional auto-encoder without contrastive loss (CAE). On a test set of 19 participants with MS with reported symptoms of COVID-19, each one paired with a participant with MS with no COVID-19 symptoms, the contrastive CAE achieves an unweighted average recall of 95 . 3% , a sensitivity of 100% and a specificity of 90 . 6% , an area under the receiver operating characteristic curve (AUC-ROC) of 0.944, indicating a maximum successful detection of symptoms in the given heart rate measurement period, whilst at the same time keeping a low false alarm rate. (c) 2021 Elsevier Ltd. All rights reserved.

KW - COVID-19

KW - Respiratory tract infection

KW - Anomaly detection

KW - Contrastive learning

KW - Convolutional auto-encoder

U2 - 10.1016/j.patcog.2021.108403

DO - 10.1016/j.patcog.2021.108403

M3 - Journal article

C2 - 34720200

VL - 123

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

M1 - 108403

ER -