Wearable sensing, big data technology for cardiovascular healthcare: current status and future prospective : Chinese Medical Journal

Secondary Logo

Journal Logo

Review Article

Wearable sensing, big data technology for cardiovascular healthcare: current status and future prospective

Miao, Fen1,2; Wu, Dan1,2; Liu, Zengding1,2; Zhang, Ruojun1,2; Tang, Min3; Li, Ye1,2

Editor(s): Wang, Ningning

Author Information
Chinese Medical Journal: September 09, 2022 - Volume - Issue - 10.1097/CM9.0000000000002117
doi: 10.1097/CM9.0000000000002117

Abstract

Introduction

Cardiovascular disease (CVD) is the leading cause of disability and death worldwide.[1,2] Approximately 17.6 million deaths were attributed to CVD in 2016, with an anticipated increase of 14.5%.[2] Previous studies showed that behavioral risk factors are major underlying causes of the CVD development. Thus, some primary cardiovascular interventions have been adopted for the cardiovascular healthcare, aiming at controlling the risk factors through lifestyle changes, such as weight reduction, smoking cessation, sleep and diet improvement, and physical activity (PA).[3] A series of guidelines or tools are widely promoted in different countries.[4,5] For example, the World Health Organization (WHO) published guidelines for assessment and management of cardiovascular risk.[6] However, traditional implementation of CVD guidelines is not satisfactory. On the one hand, it is difficult to evaluate and quantify the intervention effects in daily life, leading to a poor compliance. On the other hand, it takes a lot of energy for healthcare workers to manage long-term follow-up for high-risk groups. In the clinical setting, where the follow-up interval is usually 6 months, or even 1 year, it is difficult for medical workers to obtain the individuals’ timely physiological and behavioral parameters, such as blood pressure (BP), heart rhythm, sleep duration, and activity patterns, which are strongly associated with the progression and incidence of CVD.[7-14]

With the rapid development of wearable sensing technology in the past 15 years, an increasing number of physiological and behavioral parameters, such as heart rate (HR), PA, and sleep patterns, can be monitored by wearable devices.[15,16] With big data technology, the above parameters can be used for arrhythmia detection, hypertension diagnosis, diabetes detection, and cardiovascular risk assessment, enabling ubiquitous cardiovascular health monitoring.[16]

In this review, we first summarize the recent developments of wearable technology with promising applications in cardiovascular care. The most common wearable devices and their accuracy are highlighted. Several cardiovascular clinical applications of wearable devices are then discussed. Finally, challenges limiting the widespread adoption of wearable technology and potential solutions are highlighted.

Advances in Wearable Technology

Wearable technology has been rapidly developed in the past 20 years [Figure 1].[15] In 2006, Nike and Apple jointly launched the first wearable device recording PA.

F1
Figure 1:
The progression of wearable technology. ECG: Electrocardiogram.

Subsequently, Fitbit developed a fitness device monitoring PA, including step count, walking distance, and activity intensity, in 2008. Since 2013, many wearable products, such as the Apple Watch, Huawei wristband, and Samsung Galaxy Gear, have emerged. To adapt to different occasions, various forms of wearable devices have been developed [Table 1 presents common devices and Supplementary Table 1, https://links.lww.com/CM9/B24 presents detailed information about them], including watches or bands, patches, glasses, rings, clothes, and earphones, to collect parameters from the head, trunk, and limbs [Figure 2]. The most popular devices in the market are watches and bands, including the Apple Watch, Fitbit family, Samsung Galaxy family, and Huawei family. The primary application of wearable devices is monitoring activity patterns.[17] Recent progress in technology has enabled the tracking of various physiological parameters, such as electrocardiography (ECG) signals, photoplethysmography (PPG) signals, elec-trooculogram, electromyography, temperature, sleep, and blood oxygen. The most common parameters monitored by wearable devices are PA, heart rhythm, and sleep. A summary of the validity of various wearable devices compared with the criteria for PA, HR, and sleep is presented in Table 2 [Supplementary Table 2, https://links.lww.com/CM9/B24 presents detailed information].

Table 1 - Common wearable devices on the market.
Locations Type of device Company Product name Parameters FDA cleared Advantages/disadvantages
Head Headband Earable Earable EEG, EOG, EMG and sleep Yes Multi-parameter integration, clinical sleep scoring; less comfortable
Glass Google Google Glass HR and respiratory rate No or not sure Easy to use; few physiological parameters
Trunk Patch iRhythm iRhythm Zio HR and ECG Yes Superior clinical accuracy; may cause skin irritation and/or itching when worn
Chest strap Garmin Garmin HRM HR No or not sure HR monitor with dual transmission; less comfortable and not widely accepted
T-shirt AMSU AMSU T-shirt HR and ECG No or not sure Comfortable and lightweight; not widely accepted
Limbs Watch/band Apple Apple Watch HR, PA, sleep, and ECG Yes Wide user base and robust tracking options; expensive and low battery life
Fitbit Sense HR, PA, and sleep Yes Robust tracking options, including HR, exercise, sleep, and ECG; ECG sensors needs refining
Huawei Watch GT 2 HR, PA, and ECG Yes Good activity tracking and excellent battery; limited watch faces
BP: Blood pressure; EEG: Electroencephalogram; EOG: Electrooculogram; EMG: Electromyography; ECG: Electrocardiography; FDA: Food and Drug Administration; HR: Heart rate; PA: Physical activity; PPG: Photoplethysmography; SpO2: Blood oxygen.

F2
Figure 2:
Diverse forms of wearable devices.
Table 2 - Summary of the validity of various wearable devices compared with criteria for PA, HR, and sleep.
Parameter Reference Subjects Criterion Devices Accuracy validity
PA Case et al [27] N = 14 (4 males); 28.1 ± 6.2 years Manual counting Digi-Walker SW-200, Fitbit Zip, Fitbit One. Accuracy: The Fitbit One > Fitbit Zip > Digi-Walker SW-200.
Fokkema et al [28] N = 31 (15 males); 32 ± 12 years Manual counting Polar Loop, Garmin Vivosmart, Fitbit Charge HR, Apple Watch Sport, Pebble Smartwatch, Samsung Gear S, Misfit Flash, Jawbone Up Move, Flyfit, Moves Accuracy for slow walking: Fitbit Charge > Garmin Vivosmart; for average walking: Apple Watch > Garmin Vivosmart; for vigorous walking: Apple Watch > Samsung Gear
Xie et al [29] N = 44 (22 males): 22.2 ± 2.2 years Manual counting Apple Watch 2, Samsung Gear S3, Jawbone Up3, Fitbit Surge, Huawei Talk Band B3, Xiaomi Mi Band 2, Dongdong App, Ledongli App. Accuracy: Dongdong App > Ledongli App > Huawei Talk Band B3 > Fitbit Surge > Jawbone Up3 > Xiaomi Mi Band > Samsung Gear > Apple Watch
Boolani et al [31] [33] N = 120 (74 males); 21.4 ± 3.7 years Manual counting Fitbit Zip, Garmin Vivofit, Basis B1 band, Misfit Shine, Runtastic Pedometer, Nike FuelbandS Accuracy: Fitbit Zip > Garmin Vivofit > Basis B1 band >Misfit Shine > Nike FuelbandS > Runtastic Pedometer
Bunn et al [33] N = 20 (10 males) Manual counting Apple iWatch series 1, Fitbit Surge, Garmin 235, Moto 360, Polar A360, Suunto Spartan Sport, Suunto Spartan Trainer, and TomTom Spark 3 Accuracy during walking: Moto 360 > Garmin > Apple iWatch > Polar A360 > Suunto Trainer > TomTom > Fitbit > Suunto Sport; during running: Moto 360 > Garmin > Suunto trainer > Suunto sport > TomTom > Apple iWatch > Fitbit > Polar
HR Gillinov et al [37] N = 50 (23 males); 38 ± 12 years 12-lead ECG Polar H7, Scosche Rhythm+, Apple Watch, Fitbit Blaze, Garmin Forerunner 235, TomTom Spark Accuracy: Polar H7 > Apple Watch> TomTom Spark > Garmin Forerunner > Scosche Rhythm > Fitbit Blaze
Koshy et al [38] N = 102 (67 males); 68.0 ± 15.0 years 12-lead ECG Apple Watch Series 1 and Fitbit Blaze In sinus rhythm, both devices with a low bias (Bias = 1 beat); In atrial arrhythmias, Apple Watch (Bias = −5 beats) < Fitbit Blaze (Bias = −18 beats).
Boudreaux et al [39] N = 50 (22 males); 22.71 ± 2.99 years 6-lead ECG Apple Watch Series 2, Fitbit Blaze, Fitbit Charge 2, Polar H7, Polar A360, Garmin Vivosmart HR, TomTom Touch, and Bose SoundSport Pulse headphones Accuracy during graded exercise cycling: Apple Watch> Polar H7> Bose > TomTom > Polar A360 > Fitbit Blaze > Fitbit Charge > Garmin Vivosmart; during resistance exercise: Bose > Polar H7 > Polar A360 > Fitbit Charge > Garmin Vivosmart > Apple Watch > Fitbit Blaze >TomTom
Hwang et al [41] N = 51 (27 males); 44.1 ± 16.6 years 12-lead ECG Apple Watch Series 2, Samsung Galaxy Gear S3, and Fitbit Charge 2 The accuracy of the baseline HR measurement, that is, within ± 5 beats/min of the ECG value: Apple (100%), Galaxy (100%), and Fitbit (100%); the accuracy of the supraventricular tachyarrhythmia HR measurements to within ± 5 beats/min of the ECG value: Apple (89.3%), Galaxy (89.7%), and Fitbit (83.3%)
Sleep Baek et al [44] N = 15 males; 23.7 ± 3.0 years 12-lead ECG Polar H7 and Fitbit Charge 2 Accuracy for conventional walking and Nordic walking: Polar H7 > Fitbit Charge
De Zam6botti et al [56] N = 44 (18 males); age: 19 to 61 years PSG Fitbit Charge 2TM The bias and SD of the differences between the device and PSG for participants without PLMS: TST (−9 ± 24 min), SOL (4 ± 9 min), WASO (5 ± 19 min), time in N1+N2 (light sleep) (−34 ± 34 min), time in N3 (deep sleep) (24 ± 28 min), time in REM (1 ± 27 min).
Chinoy et al [58] N = 34 (14 males); 28.1 ± 3.9 years PSG Fatigue Science Readiband, Fitbit Alta HR, Garmin Fenix 5S, Garmin Vivosmart 3 Accuracy for sleep monitoring: Fatigue Science Readiband >Fitbit Alta HR > Garmin Vivosmart
There is a high variation in the results because of the difference in experimental protocols. Multiple studies showed that the Fitbit family of products has the highest validity for slow walking speed, while the Apple Watch is the most accurate device for average and vigorous walking speeds.
ECG patches, such as Polar H7, showed the highest accuracy. Among the selected PPG-based devices, the Apple Watch had the highest accuracy to the reference. The accuracy of PPG-based devices decreased during arrhythmias.
Wearable devices showed acceptable accuracy for TST and efficiency; however, they were inferior in detailed sleep stages. ECG: Electrocardiography; HR: Heart rate; PA: Physical activity; PLMS: Periodic limb movement of sleep; PSG: Polysomnography; REM: Rapid-eye-movement; SOL: Sleep onset latency; TST: Total sleep time; WASO: Wake after sleep onset.

PA monitoring

PA is correlated with all-cause mortality and cardiovascular outcomes including myocardial infarction, coronary heart disease, and cerebrovascular disease.[18-20] Thus, PA is a lifestyle recommendation of the WHO to promote health.[21,22] In traditional clinical research, the assessment of PA levels is based on questionnaires used during clinical visits and thus is limited by a lack of objective evaluation and by insufficiency in details such as the form and intensity of PA.[20]

Wearable devices and smartphones, which can recognize PA via an accelerometer and (or) gyroscope sensor, can objectively and accurately assess PA levels, intensity, and forms.[23] By using machine learning algorithms, such as support vector machines, neural networks, Markov chains, and Gaussian mixture models, human activities, including lying, standing, walking, running, and even the transition states, can be classified with reasonable accuracy.[24-26]

Compared with activity patterns, step count provides a useful way to quantify the activity level. Previous studies demonstrated that the risk of CVD decreased linearly with higher levels of activity.[18-20] The biological mechanisms underlying this association may be the following: PA is associated with cardiovascular risk factors such as body mass index, BP, and even inflammatory factors. Therefore, the primary function of wearable devices in PA is step count monitoring. The reliability of wearable devices in monitoring step counting has been evaluated in a few studies by comparing with direct observation[27-33]; the relative difference in mean step count ranged from 42% to −0.18% for wearable devices, indicating good accuracy for tracking step counts [Table 2]. Because of the difference in experimental protocols, there is a high variation in the results. Specifically, multiple studies have shown that among all wearable devices, the Fitbit family of products has the highest validity for a slow walking speed, with a mean absolute percent error (MAPE) <5%.[27,28,31] However, studies have also reported that the Apple Watch is the most accurate device for average and vigorous walking speeds,[28] while the Moto 360 had the least bias in step count during walking and running conditions.[33] One study reported that the Huawei Band achieved the best accuracy among six wearable devices, including the Fitbit, Jawbone Up 3, Xiaomi Band, Samsung Gear, and Apple Watch, under various physical activities including resting, walking, cycling, and sleeping.[29]

Heart rate/rhythm monitoring

Heart rates (HRs), including resting HR, maximal HR during exercise, and HR recovery following maximal exercise, are independent predictors for cardiovascular outcomes in many epidemiologic studies.[9,10,34,35] From a pathophysiological view, a relatively high HR directly affects the progression of coronary atherosclerosis, the occurrence of myocardial ischemia, and ventricular arrhythmias.[9] Continuous HR monitoring is essential for the early detection of arrhythmias and risk assessment of cardiovascular outcomes.

The gold standard for static HR is calculated as the beat-to-beat time intervals of ECG signals by a professional machine, while dynamic HR is calculated by a 24 h Holter. Electronic technologies have enabled ECG signals to be collected in various wearable forms. Wearable ECG patches attached to the chest can realize continuous heart rhythm monitoring; however, they are limited by skin intolerance and inconvenience for wearing in the long term. A single-lead ECG can be recorded from smart-watches or bands by wearing the watch on one hand and placing a finger of the other hand on the crown. Although it is suitable for measurement when necessary, this mode cannot realize continuous monitoring because two hands are required. PPG techniques can measure changes in microvascular blood volume, which can then be translated into pulse waves. The heart rhythms calculated from PPG signals are highly consistent with ECG signals.[36] Currently, PPG signals can be continuously measured by various wearable devices, including smartwatches, bands, and rings, and thus may be a good surrogate of ECG sensors for heart rhythm monitoring. The accuracy of PPG technology is limited to the degree of sensor contact with the skin, and is also influenced by skin color and moisture. In addition, the heart rhythm may not be effectively calculated under specific arrhythmia cases such as pulse shortness during arrhythmias.

The accuracy of wearable ECG and PPG sensors across different wearable devices has been compared in many studies [Table 2].[37-44] As expected, compared with wearable devices based on PPG sensors and ECG patches, Polar H7 showed the highest accuracy.[37,44] The Apple Watch had the strongest correlation or lowest MAPE among the selected PPG-based devices compared with the reference, while Fitbit devices had a higher MAPE.[37,39,44] In addition, all PPG devices tended to underestimate the HR, and this error was generally more significant at higher exercise intensities. For example, for the Fitbit Charge 2, the mean absolute error was significantly greater during Nordic walking than conventional walking (6.60 beats/ min vs. 3.68 beats/min, P < 0.001). At the same time, the reference did not differ significantly between the walking methods for Polar H7.[44] The accuracy of PPG devices decreases during arrhythmias. For example, both Apple Watch and Fitbit Blaze have a low bias (1 beat) for HR measurement in sinus rhythm; however, they increased to −5 beats for Apple Watch and −18 beats for Fitbit Blaze in atrial arrhythmias.[38] The accuracy of the baseline HR measurement within ± 5 beats/min of the reference was 100% for Apple Watch, Samsung Galaxy, and Fitbit Charge 2; however, they decreased to 89.3% for Apple Watch, 89.7% for Samsung Galaxy, and 83.3% for Fitbit.[41]

Sleep monitoring

Sleep duration has a strong correlation with cardiovascular health, with both long and short sleep duration associated with adverse outcomes, such as the high prevalence of stroke, heart failure, and diabetes.[45-47] Additionally, sleep disturbance also contributes to the risk of CVD.[48,49] Good sleep in both quality and quantity is essential for good health and thus is recommended as a lifestyle recommendation to promote health.[50] Traditionally, sleep evaluation is based on polysomnography (PSG) by attaching multiple channel electrodes to the human body, which interferes with normal sleep.

Recently, the use of signals that can be collected in a wearable way, such as ECG, PPG, and an accelerometer, has attracted widespread attention in the classification of sleep-wake stages. Data-driven techniques, such as a convolutional neural network, transfer learning, and Markov chain, have been used in this application.[51-54] Until now, validation studies comparing the accuracy of various wearable devices in sleep monitoring have been limited [Table 2].[55-58] Studies have shown good agreement between wearable devices (including Jawbone UP, Fitbit Charge 2TM,and OURA ring) and PSG for total sleep time and efficiency; however, they are inferior in the detailed sleep stages (light sleep, deep sleep, and rapid-eye-movement stage).[55-57] Conversely, the Garmin family showed worse accuracy than Fitbit devices in monitoring sleep parameters.[58]

Big Data Technology for Cardiovascular Care Using Wearable Devices

Despite rapid technological advances and widespread use, few clinical studies have evaluated wearable devices for cardiovascular care. Most studies have focused on arrhythmia detection, BP measurement, and diabetes detection [Table 3; Supplementary Table 3, https://links.lww.com/CM9/B24 presents detailed information].

Table 3 - Summary of the usage of wearable devices in cardiovascular care.
Cardiovascular care Reference Wearable device Technology Accuracy validity
Arrhythmia detection Hannun et al [61] Zio patch (iRhythm Technologies, USA) Deep neural network ROC of 0.97 to classify 12 rhythm classes on a test dataset consisting of 328 ECG records from 328 patients.
mSToPS Trial 2018 [62] Zio patch (iRhythm Technologies, USA) Cox proportional model ECG monitoring immediately resulted in a higher rate of AF diagnosis than after 4 months (3.9% vs. 0.9%). Compared with non-monitored controls, participants who received monitoring had a higher rate of AF diagnosis and greater initiation of anticoagulants at 1 year.
Apple heart study 2019 [63] Apple Watch Deep neural network Of the 86 participants who received irregular pulse notifications, 72 showed AF on concurrent ECG patch strips, with a PPV for the irregular pulse notification of 0.84.
Health eHeart Study 2018 [64] Apple Watch Deep neural network The DNN exhibited a C statistic of 0.97 to detect AF in the external validation cohort of 51 patients undergoing cardioversion. In an exploratory analysis relying on self-reported persistent AF in ambulatory participants, the C statistic was 0.72.
Bumgarner et al [65] Apple Watch combination with the AliveCor KardiaBand Compared with ECG, the Kardia Band interpreted AF with 93% sensitivity, 84% specificity, and a K coefficient of 0.77. Among 113 cases where Kardia Band and physician readings of the same recording were interpretable, the agreement was excellent (K coefficient = 0.88).
Huawei Heart Study 2019 [66] Huawei Watch GT, Honor Watch, and Honor Band PPG algorithm Of 424 participants who received a “suspected AF” notification, 227 individuals were confirmed as having AF, with the PPV of PPG signals being 91.6%.
Chen et al [67] Amazfit Health Band 1S (Huami Technology, Anhui, China) Deep learning (SEResNet) The sensitivity, specificity, and accuracy of wristband PPG readings to detect AF were 88.00%, 96.41%, and 93.27%, respectively, and those of wristband ECG readings were 87.33%, 99.20%, and 94.76%, respectively.
BP measurement Watanabe et al [70] Cuff-less BP estimation PPG algorithm The MAD between the BP value of CLB and the cuff-wearing sphygmomanometer was <8 mmHg (6.1 for SBP).
Moon et al [71] InBodyWATCH Neural network The ME was 2.2 ± 6.1 mmHg for SBP and −0.2 ± 4.2 mmHg for DBP; these were not significant (P = 0.472 for SBP and P = 0.880 for DBP). The estimated SBP/DBP ratios obtained from the InBodyWATCH within ± 5 mmHg of manual SBP/DBP were 71.4%/ 83.8%; within ± 10 mmHg, they were 86.7%/98.1%; and within ± 15 mmHg, they were 97.1%/99.0%.
Chandrasekhar et al [72] A smartphone Stepwise regression The smartphone-based device yielded bias and precision errors of 3.3 and 8.8 mmHg for SBP and −5.6 and 7.7 mmHg for DBP over a 40- to 50-mmHg range of BP.
Van Helmond et al [73] Everlast watch, BodiMetrics Statistics The average differences between the Everlast smartwatch and hospital-grade automated sphygmomanometer were systolic BP of 16.9 mmHg ± 13.5 mmHg and diastolic BP of 8.3 ± 6.1 mmHg. The average difference between the BodiMetrics performance monitor and hospital-grade automated sphygmomanometer was systolic BP of 5.3 ± 4.7 mmHg.
Diabetes, high cholesterol, etc. Ballinger et al [74] Fitbit, Apple Watch, and Wear OS Deep learning Popular wearable devices showed high accuracy at detecting multiple medical conditions, including diabetes (0.8451), high cholesterol (0.7441), high BP (0.8086), and sleep apnea (0.8298).
Diabetes detection Avram et al [75] Azumio Smartphone app Deep neural network The network achieved an area under the curve for prevalent diabetes of 0.766 in the primary cohort and 0.740 in the contemporary cohort.
Hyperkalaemia diagnosis Galloway et al [76] AliveCor ECG Deep neural network The DNN was trained to detect hyperkalemia using only ECG leads I and II. The sensitivity by duration was 94%, and the specificity was 74%.
Myocardial infarction Sopic et al [77] SmartCardia INYU Machine learning The classifier that uses all available features (n = 72) reaches a geometric mean accuracy of 83.26% (Sensitivity = 87.95%, Specificity = 78.82%).
Arterial stiffness measurement Miao et al [78] SIAT 3-in-1 Machine learning Using Omron arterial stiffness equipment as the reference, the proposed model achieved the best accuracy of 0.89, 0.2136, and 6.2432 in the correlation coefficient, ME, and standard difference for vascular age estimation.
Cardiovascular risk assessment Women's Health Study [79] ActiGraph GT3X+ Statistics (proportional hazards regression) A strong inverse association between overall volume of PA and all-cause mortality was observed. The magnitude of risk reduction (about 60%–70%) was far greater than that estimated from meta-analyses of studies using self-reported PA (about 20%–30%).
Akbulut et al [80] CVDiMo Machine learning The use of PA results and stress levels deduced from the emotional state analysis resulted in better risk estimation. The highest accuracy of classifying the short-term health status was 96%.
ECG patches can classify multiple rhythms with acceptable accuracy, while PPG-based devices showed acceptable accuracy in detecting AF.
Wearable devices showed acceptable accuracy in controlled settings, the validation in large-scale population is rare.
Research on the use of wearable devices in detecting or predicting CVDs is rare; preliminary studies showed the potential of wearable devices in cardiovascular healthcare. AF: Atrial fibrillation; BP: Blood pressure; CVD: Cardiovascular disease; CVDiMo: Cardiovascular Disease Monitoring; CLB: Cuff-less BP estimation; DBP: Diastolic blood pressure; DNN: Deep neural network; ECG: Electrocardiogram; MAD: Mean absolute difference; ME: Mean error; PA: Physical activity; PPG: Photoplethysmography; PPV: Positive predictive value; ROC: Receiver operating characteristic curve; SBP: Systolic blood pressure.

Atrial fibrillation (AF) and other arrhythmia detections

The global burden of AF has become a worldwide concern. The worldwide prevalence of AF is 37,574 million cases (0.51% of the worldwide population), which increased by 33% over the last 20 years.[59] As with the rate or rhythm of the heartbeat, AF and other arrhythmias can be well identified from ECG signals. Because of the advantage of wearable devices in monitoring HR, the detection of arrhythmia, especially AF, is a promising application of wearable devices.[60] In 2019, a deep neural network (DNN) was trained on 91,232 single-lead ECGs to classify 12 rhythm classes using the Zio monitor (iRhythm Technologies, USA).[61] The DNN achieved an average area under the receiver operating characteristic curve (ROC) of 0.97 on a test dataset consisting of 328 ECG records from 328 patients. The Zio monitor was also evaluated in the mSToPS study to facilitate new AF diagnoses.[62] This study showed that ECG monitoring immediately resulted in a higher rate of AF diagnosis than after 4 months (3.9% vs. 0.9%). Compared with non-monitored controls, participants who received monitoring had a higher rate of AF diagnosis and greater initiation of anticoagulants at 1 year. Wristbands or smartwatches were also evaluated to identify arrhythmias based on PPG signals. Using the Apple Watch-based PPG data, 419,297 participants were enrolled in the Apple Heart study to determine whether the device could detect AF. Among the 2161 participants with irregular pulse notification, 34% of the incidents were confirmed as AF using an ECG patch. The positive predictive value (PPV) for AF detection was 84% for irregular notification algorithms.[63] In another study using Apple Watch PPG data, a DNN was trained from 9750 participants (including 347 participants with AF) to detect AFs. The DNN exhibited a C statistic of 0.97 (95% confidence interval [CI]: 0.94–1.00; P < 0.001) in the detection of AF against the reference standard 12-lead ECG in the external validation cohort of 51 patients.[64] Combined with the Apple Watch, the Kardia Band interpreted AF with a sensitivity of 93% and specificity of 84% compared with physician interpretation based on 12-lead ECGs.[65] The Huawei Heart Study aimed to determine the feasibility of AF screening in a large population cohort using PPG data from the Huawei Watch/Band. Among those wearing wearable devices, 424 received a “suspected AF” notification (424/187,912, 0.23%). Among those effectively followed up, 227 individuals (227/262, 87.0%) were confirmed as having AF, with the PPV of PPG signals being 91.6%.[66] The Amazfit Health Band 1S equipped with ECG and PPG sensors from Huami was also evaluated in 2020 to detect AF using deep learning technology. The proposed network achieved a sensitivity, specificity, and accuracy of 88.00%, 96.41%, and 93.27%, respectively, for wristband PPG, and 87.33%, 99.20%, and 94.76%, respectively, for wristband ECG readings.[67]

BP measurement

Clinical studies have reported that BP and its variability are associated with CVD or mortality.[68,69] Oscillometric-based technologies are mostly used for BP measurement; however, they are characterized by discomfort and disruption to daily life and sleep, particularly for ambulatory BP monitoring. Cuff-less wearables that can be continuously worn are a promising technique for screening hypertension and evaluating BP variability in the ambulatory setting. Currently, wearable devices for BP measurement focus on BP estimation using ECG signals and/or PPG signals using pulse transit time based or machine learning based methods.[69] A cuff-less BP estimation system based on one PPG sensor was reported.[70] The system's performance meets the latest wearable device standard (issued by the IEEE standard 1708–2014). The system is expected to offer a flexible and wearable device that permits BP monitoring in more continuous and stress-free settings. In 2020, a neural network was proposed to estimate BP based on a wearable device, InBodyWATCH. The ratio of estimated systolic blood pressure (SBP) to diastolic blood pressure (DBP) obtained from InBodyWATCH is 71.4%/83.8% in the ± 5 mmHg range of manual SBP/DBP, 86.7%/98.1% within ± 10 mmHg, and 97.1%/99.0% within ± 15 mmHg, showing good accuracy that suggests it may be used for dynamic BP monitoring.[71] Smartphone-based BP monitoring via the oscillometric finger-pressing method was proposed in 2018, involving a PPG and a force sensor unit. The smartphone-based device yielded bias and precision errors of 3.3 and 8.8 mmHg for SBP and −5.6 and 7.7 mmHg for DBP over a 40 to 50 mmHg range of BP, verifying the feasibility of cuff-less and calibration-free monitoring via a smartphone.[72] The above methods for BP measurement are still at the experimental laboratory stage. The accuracy of two commercial products, including the Everlast watch and BodiMetrics, was compared.[73] The result showed that both presented low accuracy and were not competitive as BP measurement devices. The average differences between the Everlast smartwatch and hospital-grade automated sphygmomanometer for SBP were 16.9 ± 13.5 mmHg and 8.3 ± 6.1 mmHg for DBP. The average difference between the BodiMetrics and hospital-grade automated sphygmomanometer was 5.3 ± 4.7 mmHg for SBP.[73] Therefore, the accuracy of the cuff-less BP measurement using wearable devices is still controversial because of the lack of a large-scale cohort assessment. Further improvements are needed to use commercial devices to monitor BP.

Other diagnostic applications

Wearable devices have also been used in cardiovascular care settings, including diabetes detection, hyperkalemia diagnosis, myocardial infarction detection, and arterial stiffness evaluation. In 2018, a multi-task long-short term machine based method was proposed to detect diabetes, high cholesterol, high BP, and sleep apnea using the Fitbit, Apple Watch, and Wear OS wearable devices. The results showed high accuracy at detecting multiple cardiovascular abnormalities, including diabetes (0.8451), high cholesterol (0.7441), high BP (0.8086), and sleep apnea (0.8298).[74] In 2020, a DNN was trained to detect diabetes on 53,879 individuals based on PPG data collected from smartphones. The network achieved an ROC of 0.766 for prevalent diabetes in a contemporary cohort of 7806 (95% CI: 0.750–0.782; sensitivity 75%, specificity 65%) and 0.740 in a clinic cohort of 181 (95% CI: 0.723–0.758; sensitivity 81%, specificity 54%).[75] The utility of AliverCor ECG for hyperkalemia diagnosis was proposed based on deep learning techniques. The proposed algorithm achieved a sensitivity of 94% and specificity of 74% in hyperkalemia diagnosis for patients wearing AliveCor ECG.[76] In 2018, machine learning algorithms were employed for early detection and prevention of myocardial infarction using the ECG data collected by the SmartCardia INYU device. Based on random forest classifications, the proposed method classified myocardial infarction with an accuracy of 83.26% (sensitivity 87.95%, specificity 78.82%).[77] A wearable device equipped with an ECG and PPG sensor, SIAT 3-in-1, was used to evaluate arterial stiffness based on machine learning algorithms. By using Omron arterial stiffness equipment as the reference, the proposed model achieved the best accuracy of 0.89, 0.2136, and 6.2432 in terms of the correlation coefficient, mean difference, and standard difference for vascular age estimation.[78]

Cardiovascular risk assessment

Until now, studies using wearable devices for long-term cardiovascular risk prediction are very limited. The Women's Health Study validated the association between PA collected by ActiGraph GT3X+ and all-cause mortality. Results showed a strong inverse association between the overall volume of PA and all-cause mortality. The magnitude of risk reduction (about 60%–70%, comparing extreme quartiles) was far greater than that estimated from meta-analyses of studies using self-reported PA (about 20%–30%).[79] A wearable device (Cardiovascular Disease Monitoring) for continuous monitoring of CVDs was proposed in 2018. Six different biological signals involved in two different test groups were analyzed and tested. In addition to examining the patient's biosignals, the use of PA results and stress levels inferred from the emotional state analysis achieved higher performance in risk estimation. The highest accuracy for determining short-term health status was 96% in the study.[80]

Challenges and Future Directions

Wearable devices can provide continuous health care monitoring without interference or disruption to daily life and thus are promising for long-term cardiovascular care. First, these devices can aid the early detection of asymptomatic cardiovascular abnormalities such as arrhythmias. Second, wearable devices enable dynamic assessment of the development trends of CVD by providing continuous monitoring of parameters. Third, wearable devices are helpful in the timely detection of acute cardiovascular events by identifying the abnormalities of health parameters such as ventricular arrhythmias or cardiac arrest. Finally, wearable devices enable effective self-management interventions for the users, including modifications in sleep, exercise, and medications, by helping them pay more attention to their health parameters. Even though wearable devices have been used in diverse applications and provide knowledge for cardiovascular healthcare, several challenges are still hindering their widespread applications in clinical practice.

Accuracy and validity of wearable devices

To ensure the reliability of clinical studies aiming to assess the effectiveness of wearable devices in detecting cardiovascular anomalies, an indispensable characteristic for such devices is their high accuracy in monitoring the physiological and behavior parameters of the users. The signals collected by wearable devices may be largely corrupted because of the body movement, contact degree with the skin, noise, and motion artifacts during daily life. The reliability of these devices is dependent on the algorithms designed for processing the raw data. Signal quality assessment algorithms can help extract reliable signals from the raw data to evaluate and diagnose the users’ cardiovascular health. Although numerous algorithms adopting signal processing or machine learning for signal quality assessment have been proposed,[81,82] the high variability of physiological signals among different populations should be emphasized. Another challenge is that physiological signals under normal health status can vary significantly, thus increasing the difficulty of discerning abnormal conditions from normal conditions. With the help of states information identified from other sensors, context-aware methods should be considered for an adaptive signal quality assessment.

Many studies have been conducted to assess the accuracy of wearable technology; however, because of the lack of standards and experimental protocols, wide heterogeneity is observed across different studies, and even some validation studies have questioned the accuracy of wearable technology.[83-85] Additionally, an increased number of wearable devices have flooded the market without sufficient validation, leading to the public's anti-dependence attitude. Fortunately, at the beginning of 2020, Coravos et al[86] developed an evaluation framework to test the accuracy and validity of connected sensor technologies, which included wearables. In addition to the framework, we suggest that the authorities should develop standard validation protocols to allow for a fair comparison of accuracy among wearable devices.

Data redundancy

The data collected from wearable devices has high redundancy and low value because of long-term monitoring under non-standard conditions. Continuous wearing of devices will generate a large amount of physiological data, with many data being repeated or corrupted, particularly for healthy users and during motion states. From the perspective of the mechanism, the health status in a short period is stable for a healthy population, which leads to redundant physiological information (day-by-day) from wearable devices. In addition, the same piece of data exists from multiple wearables; for example, both smartwatches and smartphones generate accelerometer data. Such a situation causes data inconsistency between devices, providing unreliable and/or meaningless information. Coupled with the corrupted data caused by noise and motion artifacts in daily life, wearable devices generate a large amount of redundant and low-value data. The means to process and store these data has become a great challenge. Incremental data storage and processing frameworks should be considered to avoid redundant low-value data storage.[87]

Lack of criteria with clinical evidence

Clinical examination is traditionally required to follow a standard process; for example, the BP measurement should be performed in a quiet state and at the same horizontal level between the device and heart. However, the data acquisition scenario of wearable devices is very complex; for example, the BP may be collected under a motion state. In fact, there are no clinical criteria of health parameters under multiple states. The means by which the relationship between wearable data and clinical criteria is constructed is also a significant challenge. Large-scale prospective cohort studies should be conducted to build the relationship between wearable data and cardiovascular outcomes.

Data security and management

Because of the sensitivity of wearable data, data privacy is a major issue that should be emphasized in the big health era. Currently, most wearable data are only available to device manufacturers, with no effective supervision and no clear definition of ownership. With the rapid increase of users, how do you protect sensitive wearable data from data leakage? How do you avoid data tampering or improper use of wearable data? How do you facilitate data sharing for clinical research? Therefore, it is necessary for the authorities to build data security and sharing architecture among certified wearable devices. In addition, data security methods, such as blockchain,[88] should be considered to ensure the constancy of wearable data during data sharing.

Miniaturization of wearable devices

Although wearable devices are widely used, the miniaturization of devices should be further studied to improve the compliance of the potential users. The collection of vital signs for cardiovascular healthcare should be embedded into the articles that users wear daily, without additional burden. Therefore, the technology for hardware and system miniaturization, such as a medical chip for physiological signals’ collection, flexible electronics materials, and energy harvesting,[89,90] need to be adopted in wearable devices to improve the portability and usability.

Future research directions

Even though challenges exist, wearable technology still provides promising opportunities in long-term or even life-cycle cardiovascular care by containing abundant information on a small-time scale. To apply wearable devices to clinical practice, the device accuracy and usability should first be verified through large-scale clinical trials that follow standard experimental protocols. After full validation, a large-scale prospective cohort study integrating clinical and wearable data should be conducted to build the relationship between wearable data and long-term cardiovascular outcomes such as heart failure and coronary heart diseases. Through this type of research, independent novel risk factors, such as the variability of physiological parameters, may be identified from continuous wearable data. Perhaps indicators extracted from wearable devices have no clear clinical implications at this time, however, they may still be related to cardiovascular outcomes because of the time-varied information. An urgent clinical application of wearable devices is monitoring and predicting acute cardiovascular events such as myocardial infarction and cardiac arrest. Typical characteristics before acute events might be identified through continuous monitoring. The main challenge of this application is the means to capture a large number of acute events for modeling. With the development of wearable, big data technology, and clinical research, wearable devices would enable a new era for cardiovascular healthcare.

Conclusions

Wearable technology has developed rapidly in the past 20 years. Physiological and behavioral parameters can be continuously monitored by wearable devices with reasonable accuracy. With the help of big data technology, wearable devices have been applied in various cardiovascular care settings, such as early detection of arrhythmias, BP measurement, and diabetes diagnosis. They have also demonstrated great promise in cardiovascular risk assessment. However, some challenges still exist for the widespread application of wearable devices. Validation studies following standard experimental protocols should be conducted to enable the use of wearable devices that can provide high-quality and reliable cardiovascular health care. In addition, wearable data must be supervised by effective regulatory policies to ensure data security. In the future, a large-scale prospective cohort study integrating clinical and wearable data should be conducted to construct the relationship between wearable data and long-term cardiovascular outcomes or acute cardiovascular events.

Funding

This study was supported in part by the National Natural Science Foundation of China (No. U1913210), in part by the Strategic Priority CAS Project (XDB38040200), and in part by the Basic Research Project of Shenzhen (JCYJ20210324101206017).

Conflicts of interest

None.

References

1. Timmis A, Townsend N, Gale C, Grobbee R, Maniadakis N, Flather M, et al. European society of cardiology: cardiovascular disease statistics 2017. Eur Heart J 2018;39:508–579. doi: 10.1093/eurheartj/ehx628.
2. Benjamin EJ, Muntner P, Alonso A, Bittencourt MS, Callaway CW, Carson AP, et al. Heart disease and stroke statistics-2019 update: a report from the American Heart Association. Circulation 2019;139:e56–e528. doi: 10.1161/CIR.0000000000000659.
3. Nilsson PM, Klasson EB, Nyberg P. Life-style intervention at the worksite - reduction of cardiovascular risk factors in a randomized study. Scand J Work Environ Health 2001;27:57–62. doi: 10.5271/sjweh.587.
4. Ferket BS, Colkesen EB, Visser JJ, Spronk S, Kraaijenhagen RA, Steyerberg EW, et al. Systematic review of guidelines on cardiovascular risk assessment: which recommendations should clinicians follow for a cardiovascular health check? Arch Intern Med 2010;170:27–40. doi: 10.1001/archinternmed.2009.434.
5. McGorrian C, Yusuf S, Islam S, Jung H, Rangarajan S, Avezum A, et al. Estimating modifiable coronary heart disease risk in multiple regions of the world: the INTERHEART modifiable risk score. Eur Heart J 2011;32:581–589. doi: 10.1093/eurheartj/ehq448.
6. Organization WH. Prevention of cardiovascular disease: guidelines for assessment and management of total cardiovascular risk. Geneva(Switzerland): World Health Organization; 2007.
7. Vasan RS, Larson MG, Leip EP, Evans JC, O’Donnell CJ, Kannel WB, et al. Impact of high-normal blood pressure on the risk of cardiovascular disease. N Engl J Med 2001;345:1291–1297. doi: 10.1056/NEJMoa003417.
8. Kshirsagar AV, Carpenter M, Bang H, Wyatt SB, Colindres RE. Blood pressure usually considered normal is associated with an elevated risk of cardiovascular disease. Am J Med 2006;119:133–141. doi: 10.1016/j.amjmed.2005.08.023.
9. Fox K, Borer JS, Camm AJ, Danchin N, Ferrari R, Sendon JLL, et al. Resting heart rate in cardiovascular disease. J Am Coll Cardiol 2007;50:823–830. doi: 10.1016/j.jacc.2007.04.079.
10. Thayer JF, Yamamoto SS, Brosschot JF. The relationship of autonomic imbalance, heart rate variability and cardiovascular disease risk factors. Int J Cardiol 2010;141:122–131. doi: 10.1016/j. ijcard.2009.09.543.
11. Ignarro LJ, Balestrieri ML, Napoli C. Nutrition, physical activity, and cardiovascular disease: an update. Cardiovasc Res 2007;73:326–340. doi: 10.1016/j.cardiores.2006.06.030.
12. Carnethon MR. Physical activity and cardiovascular disease: how much is enough? Am J Lifestyle Med 2009;3:44S–49S. doi: 10.1177/1559827609332737.
13. Cappuccio FP, Cooper D, D’Elia L, Strazzullo P, Miller MA. Sleep duration predicts cardiovascular outcomes: a systematic review and meta-analysis of prospective studies. Eur Heart J 2011;32:1484–1492. doi: 10.1093/eurheartj/ehr007.
14. Reference Values for Arterial Stiffness’ Collaboration. Determinants of pulse wave velocity in healthy people and in the presence of cardiovascular risk factors: “Establishing normal and reference values”. Eur Heart J 2010;31:2338–2350. doi: 10.1093/eurheartj/ehq165.
15. Iqbal MH, Aydin A, Brunckhorst O, Dasgupta P, Ahmed K. A review of wearable technology in medicine. J R Soc Med 2016;109:372–380. doi: 10.1177/0141076816663560.
16. Bayoumy K, Gaber M, Elshafeey A, Mhaimeed O, Dineen EH, Marvel FA, et al. Smart wearable devices in cardiovascular care: where we are and how to move forward. Nat Rev Cardiol 2021;18:581–599. doi: 10.1038/s41569-021-00522-7.
17. Strain T, Wijndaele K, Dempsey PC, Sharp SJ, Pearce M, Jeon J, et al. Wearable-device-measured physical activity and future health risk. Nat Med 2020;26:1385–1391. doi: 10.1038/s41591-020-1012-3.
18. Yusuf S, Hawken S, Ounpuu S, Dans T, Avezum A, Lanas F, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet 2004;364:937–952. doi: 10.1016/S0140-6736(04)17018-9.
19. Paffenbarger RS Jr, Blair SN, Lee IM. A history of physical activity, cardiovascular health and longevity: the scientific contributions of Jeremy N Morris, DSc, DPH, FRCP. Int J Epidemiol 2001;30:1184–1192. doi: 10.1093/ije/30.5.1184.
20. Williams MA, Haskell WL, Ades PA, Amsterdam EA, Bittner V, Franklin BA, et al. Resistance exercise in individuals with and without cardiovascular disease: 2007 update: a scientific statement from the American Heart Association Council on Clinical Cardiology and Council on Nutrition, Physical Activity, and Metabolism. Circulation 2007;116:572–584. doi: 10.1161/CIRCULATIO-NAHA.107.185214.
21. World Health Organization T. Global Recommendations on Physical Activity for Health. World Health Organization 2010. Available from: https://apps.who.int/iris/bitstream/handle/10665/44399/9789245599975_chi.pdf. [Accessed on December 1, 2021]
22. Haskell WL, Lee IM, Pate RR, Powell KE, Blair SN, Franklin BA, et al. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Circulation 2007;116:1081–1093. doi: 10.1161/CIRCULATIONAHA.107.185649.
23. Najafi B, Aminian K, Paraschiv-Ionescu A, Loew F, Büla CJ, Robert P. Ambulatory system for human motion analysis using a kinematic sensor: monitoring of daily physical activity in the elderly. IEEE Trans Biomed Eng 2003;50:711–723. doi: 10.1109/TBME.2003.812189.
24. Ermes M, Pärkka J, Mantyjarvi J, Korhonen I. Detection of daily activities and sports with wearable sensors in controlled and uncontrolled conditions. IEEE Trans Inf Technol Biomed 2008;12:20–26. doi: 10.1109/TITB.2007.899496.
25. Allen FR, Ambikairajah E, Lovell NH, Celler BG. Classification of a known sequence of motions and postures from accelerometry data using adapted Gaussian mixture models. Physiol Meas 2006;27:935–951. doi: 10.1088/0967-3334/27/10/001.
26. Kern N, Schiele B, Junker H, Lukowicz P, Troster G. Wearable sensing to annotate meeting recordings. Proc Sixth Int Symp Wear Comput, IEEE 2002;186–193. doi: 10.1109/ISWC.2002.1167247.
27. Case MA, Burwick HA, Volpp KG, Patel MS. Accuracy of smartphone applications and wearable devices for tracking physical activity data. JAMA 2015;313:625–626. doi: 10.1001/jama.2014.17841.
28. Fokkema T, Kooiman TJM, Krijnen WP, Van der Schans CP, De Groot M. Reliability and validity of ten consumer activity trackers depend on walking speed. Med Sci Sports Exerc 2017;49:793–800. doi: 10.1249/MSS.0000000000001146.
29. Xie J, Wen D, Liang L, Jia Y, Gao L, Lei J. Evaluating the validity of current mainstream wearable devices in fitness tracking under various physical activities: comparative study. JMIR Mhealth Uhealth 2018;6:e94. doi: 10.2196/mhealth.9754.
30. Höchsmann C, Knaier R, Eymann J, Hintermann J, Infanger D, Schmidt-Trucksäss A. Validity of activity trackers, smartphones, and phone applications to measure steps in various walking conditions. Scand J Med Sci Sports 2018;28:1818–1827. doi: 10.1111/sms.13074.
31. Boolani A, Towler C, LeCours B, Blank H, Larue J, Fulk G. Accuracy of 6 commercially available activity monitors in measuring heart rate, caloric expenditure, steps walked, and distance traveled. Cardiopulm Phys Ther J 2019;30:153–161. doi: 10.1097/CPT.0000000000000107.
32. Navalta JW, Montes J, Bodell NG, Aguilar CD, Lujan A, Guzman G, et al. Wearable device validity in determining step count during hiking and trail running. J Measure Phys Behav 2018;1:86–93. doi: 10.1123/jmpb.2018-0010.
33. Bunn JA, Jones C, Oliviera A, Webster MJ. Assessment of step accuracy using the Consumer Technology Association standard. J Sports Sci 2019;37:244–248. doi: 10.1080/02640414.2018.1491941.
34. Morshedi-Meibodi A, Larson MG, Levy D, O’Donnell CJ, Vasan RS. Heart rate recovery after treadmill exercise testing and risk of cardiovascular disease events (The Framingham Heart Study). Am J Cardiol 2002;90:848–852. doi: 10.1016/s0002-9149(02)02706-6.
35. Cole CR, Foody JM, Blackstone EH, Lauer MS. Heart rate recovery after submaximal exercise testing as a predictor of mortality in a cardiovascularly healthy cohort. Ann Intern Med 2000;132:552–555. doi: 10.7326/0003-4819-132-7-200004040-00007.
36. Jan HY, Chen MF, Fu TC, Lin WC, Tsai CL, Lin KP. Evaluation of coherence between ECG and PPG derived parameters on heart rate variability and respiration in healthy volunteers with/without controlled breathing. J Med Biol Eng 2019;39:783–795. doi: 10.1007/s40846-019-00468-9.
37. Gillinov S, Etiwy M, Wang R, Blackburn G, Phelan D, Gillinov AM, et al. Variable accuracy of wearable heart rate monitors during aerobic exercise. Med Sci Sports Exerc 2017;49:1697–1703. doi: 10.1249/MSS.0000000000001284.
38. Koshy AN, Sajeev JK, Nerlekar N, Brown AJ, Rajakariar K, Zureik M, et al. Smart watches for heart rate assessment in atrial arrhythmias. Int J Cardiol 2018;266:124–127. doi: 10.1016/j. ijcard.2018.02.073.
39. Boudreaux BD, Hebert EP, Hollander DB, Williams BM, Cormier CL, Naquin MR, et al. Validity of wearable activity monitors during cycling and resistance exercise. Med Sci Sports Exerc 2018;50:624–633. doi: 10.1249/MSS.0000000000001471.
40. Thiebaud RS, Funk MD, Patton JC, Massey BL, Shay TE, Schmidt MG, et al. Validity of wrist-worn consumer products to measure heart rate and energy expenditure. Digit Health 2018;4:2055207618770322. doi: 10.1177/2055207618770322.
41. Hwang J, Kim J, Choi KJ, Cho MS, Nam GB, Kim YH. Assessing accuracy of wrist-worn wearable devices in measurement of paroxysmal supraventricular tachycardia heart rate. Korean Circ J 2019;49:437–445. doi: 10.4070/kcj.2018.0323.
42. Thomson EA, Nuss K, Comstock A, Reinwald S, Blake S, Pimentel RE, et al. Heart rate measures from the Apple Watch, Fitbit Charge HR 2, and electrocardiogram across different exercise intensities. J Sports Sci 2019;37:1411–1419. doi: 10.1080/02640414.2018.1560644.
43. Navalta JW, Montes J, Bodell NG, Salatto RW, Manning JW, DeBeliso M. Concurrent heart rate validity of wearable technology devices during trail running. PLoS One 2020;15:e0238569. doi: 10.1371/journal.pone.0238569.
44. Baek S, Ha Y, Park HW. Accuracy of wearable devices for measuring heart rate during conventional and nordic walking. PM R 2021;13:379–386. doi: 10.1002/pmrj.12424.
45. Cash RE, Hery CMB, Panchal AR, Bower JK. Association between sleep duration and ideal cardiovascular health among US adults, National Health and Nutrition Examination Survey, 2013-2016. Prev Chronic Dis 2020;17:E43. doi: 10.5888/pcd17.190424.
46. Krittanawong C, Kumar A, Wang Z, Jneid H, Baber U, Mehran R, et al. Sleep duration and cardiovascular health in a representative community population (from NHANES, 2005 to 2016). Am J Cardiol 2020;127:149–155. doi: 10.1016/j.amjcard.2020.04.012.
47. Jackson CL, Redline S, Emmons KM. Sleep as a potential fundamental contributor to disparities in cardiovascular health. Annu Rev Public Health 2015;36:417–440. doi: 10.1146/annurev-publhealth-031914-122838.
48. Grandner MA, Jackson NJ, Pak VM, Gehrman PR. Sleep disturbance is associated with cardiovascular and metabolic disorders. J Sleep Res 2012;21:427–433. doi: 10.1111/j.1365-2869.2011.00990.x.
49. Narang I, Manlhiot C, Davies-Shaw J, Gibson D, Chahal N, Stearne K, et al. Sleep disturbance and cardiovascular risk in adolescents. CMAJ 2012;184:E913–E920. doi: 10.1503/cmaj.111589.
50. Mukherjee S, Patel SR, Kales SN, Ayas NT, Strohl KP, Gozal D, et al. An official American Thoracic Society Statement: the importance of healthy sleep. Recommendations and future priorities. Am J Respir Crit Care Med 2015;191:1450–1458. doi:10.1164/rccm.201504-0767ST.
51. Walch O, Huang Y, Forger D, Goldstein C. Sleep stage prediction with raw acceleration and photoplethysmography heart rate data derived from a consumer wearable device. Sleep 2019;42:zsz180. doi: 10.1093/sleep/zsz180.
52. Cakmak AS, Da Poian G, Willats A, Haffar A, Abdulbaki R, Ko YA, et al. An unbiased, efficient sleep-wake detection algorithm for a population with sleep disorders: change point decoder. Sleep 2020;43:zsaa011. doi:10.1093/slee/zsaa011.
53. Li Q, Li Q, Liu C, Shashikumar SP, Nemati S, Clifford GD. Deep learning in the cross-time frequency domain for sleep staging from a single-lead electrocardiogram. Physiol Meas 2018;39:124005. doi: 10.1088/1361-6579/aaf339.
54. Li Q, Li Q, Cakmak AS, Da Poian G, Bliwise DL, Vaccarino V, et al. Transfer learning from ECG to PPG for improved sleep staging from wrist-worn wearables. Physiol Meas 2021;42:044004. doi:10.1088/1361-6579/abf1b0.
55. de Zambotti M, Baker FC, Colrain IM. Validation of sleep-tracking technology compared with polysomnography in adolescents. Sleep 2015;38:1461–1468. doi: 10.5665/sleep.4990.
56. de Zambotti M, Goldstone A, Claudatos S, Colrain IM, Baker FC. A validation study of Fitbit Charge 2TM compared with polysomnog-raphy in adults. Chronobiol Int 2018;35:465–476. doi: 10.1080/07420528.2017.1413578.
57. de Zambotti M, Rosas L, Colrain IM, Baker FC. The sleep of the ring: comparison of the ✉URA sleep tracker against polysomnography. Behav Sleep Med 2019;17:124–136. doi: 10.1080/15402002.2017.1300587.
58. Chinoy ED, Cuellar JA, Huwa KE, Jameson JT, Watson CH, Bessman SC, et al. Performance of seven consumer sleep-tracking devices compared with polysomnography. Sleep 2021;44:zsaa291. doi: 10.1093/sleep/zsaa291.
59. Lippi G, Sanchis-Gomar F, Cervellin G. Global epidemiology of atrial fibrillation: an increasing epidemic and public health challenge. Int J Stroke 2021;16:217–221. doi: 10.1177/1747493019897870.
60. Ip JE. Wearable devices for cardiac rhythm diagnosis and management. JAMA 2019;321:337–338. doi: 10.1001/jama.2018.20437.
61. Hannun AY, Rajpurkar P, Haghpanahi M, Tison GH, Bourn C, Turakhia MP, et al. Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network. Nat Med 2019;25:65–69. doi: 10.1038/s41591-018-0268-3.
62. Steinhubl SR, Waalen J, Edwards AM, Ariniello LM, Mehta RR, Ebner GS, et al. Effect of a home-based wearable continuous ECG monitoring patch on detection of undiagnosed atrial fibrillation: the mSToPS randomized clinical trial. JAMA 2018;320:146–155. doi: 10.1001/jama.2018.8102.
63. Perez MV, Mahaffey KW, Hedlin H, Rumsfeld JS, Garcia A, Ferris T, et al. Large-scale assessment of a smartwatch to identify atrial fibrillation. N Engl J Med 2019;381:1909–1917. doi: 10.1056/NEJMoa1901183.
64. Tison GH, Sanchez JM, Ballinger B, Singh A, Olgin JE, Pletcher MJ, et al. Passive detection of atrial fibrillation using a commercially available smartwatch. JAMA Cardiol 2018;3:409–416. doi: 10.1001/jamacardio.2018.0136.
65. Bumgarner JM, Lambert CT, Hussein AA, Cantillon DJ, Baranowski B, Wolski K, et al. Smartwatch algorithm for automated detection of atrial fibrillation. J Am Coll Cardiol 2018;71:2381–2388. doi: 10.1016/j.jacc.2018.03.003.
66. Guo Y, Wang H, Zhang H, Liu T, Liang Z, Xia Y, et al. Mobile photoplethysmographic technology to detect atrial fibrillation. J Am Coll Cardiol 2019;74:2365–2375. doi: 10.1016/j.jacc.2019.08.019.
67. Chen E, Jiang J, Su R, Gao M, Zhu S, Zhou J, et al. A new smart wristband equipped with an artificial intelligence algorithm to detect atrial fibrillation. Heart Rhythm 2020;17:847–853. doi: 10.1016/j. hrthm.2020.01.034.
68. Stevens SL, Wood S, Koshiaris C, Law K, Glasziou P, Stevens RJ, et al. Blood pressure variability and cardiovascular disease: systematic review and meta-analysis. BMJ 2016;354:i4098. doi: 10.1136/bmj.i4098.
69. Ding XR, Zhao N, Yang GZ, Pettigrew RI, Lo B, Miao F, et al. Continuous blood pressure measurement from invasive to unobtrusive: celebration of 200th birth anniversary of Carl Ludwig. IEEE J Biomed Health Inform 2016;20:1455–1465. doi: 10.1109/JBHI.2016.2620995.
70. Watanabe N, Bando YK, Kawachi T, Yamakita H, Futatsuyama K, Honda Y, et al. Development and validation of a novel cuff-less blood pressure monitoring device. JACC Basic Transl Sci 2017;2:631–642. doi: 10.1016/j.jacbts.2017.07.015.
71. Moon JH, Kang MK, Choi CE, Min J, Lee HY, Lim S. Validation of a wearable cuff-less wristwatch-type blood pressure monitoring device. Sci Rep 2020;10:19015. doi: 10.1038/s41598-020-75892-y.
72. Chandrasekhar A, Kim CS, Naji M, Natarajan K, Hahn JO, Mukkamala R. Smartphone-based blood pressure monitoring via the oscillometric finger-pressing method. Sci Transl Med 2018;10:eaa8674. doi: 10.1126/scitranslmed.aap8674.
73. van Helmond N, Freeman CG, Hahnen C, Haldar N, Hamati JN, Bard DM, et al. The accuracy of blood pressure measurement by a smartwatch and a portable health device. Hosp Pract 1995- 2019;47:211–215. doi: 10.1080/21548331.2019.1656991.
74. Ballinger B, Hsieh J, Singh A, Sohoni N, Wang J, Tison G, et al. DeepHeart: semi-supervised sequence learning for cardiovascular risk prediction. Proc AAAI Conf Artif Intell 2018;2079–22086. Available from: https://ojs.aaai.org/index.php/AAAI/article/view/11891.
75. Avram R, Olgin JE, Kuhar P, Hughes JW, Marcus GM, Pletcher MJ, et al. A digital biomarker of diabetes from smartphone-based vascular signals. Nat Med 2020;26:1576–1582. doi: 10.1038/s41591-020-1010-5.
76. Galloway CD, Valys AV, Petterson FL, Gundotra VP, Treiman DL, Albert DE, et al. Non-invasive detection of hyperkalemia with a smartphone electrocardiogram and artificial intelligence. J Am Coll Cardiol 2018;71:A272. doi: 10.1016/S0735-1097(18)30813-1.
77. Sopic D, Aminifar A, Aminifar A, Atienza D. Real-time event-driven classification technique for early detection and prevention of myocardial infarction on wearable systems. IEEE Trans Biomed Circuits Syst 2018;12:982–992. doi: 10.1109/TBCAS.2018.2848477.
78. Miao F, Wang X, Yin L, Li Y. A wearable sensor for arterial stiffness monitoring based on machine learning algorithms. IEEE Sens J 2019;19:1426–1434. doi: 10.1109/JSEN.2018.2880434.
79. Lee IM, Shiroma EJ, Evenson KR, Kamada M, LaCroix AZ, Buring JE. Accelerometer-measured physical activity and sedentary behavior in relation to all-cause mortality: The Women's Health Study. Circulation 2018;137:203–205. doi: 10.1161/CIRCULATIO-NAHA.117.031300.
80. Akbulut FP, Akan A. A smart wearable system for short-term cardiovascular risk assessment with emotional dynamics. Measurement 2018;128:237–246. doi: 10.1016/j.measurement.2018.06.050.
81. Satija U, Ramkumar B, Manikandan MS. A review of signal processing techniques for electrocardiogram signal quality assessment. IEEE Rev Biomed Eng 2018;11:36–52. doi: 10.1109/RBME.2018.2810957.
82. Li Q, Clifford GD. Dynamic time warping and machine learning for signal quality assessment of pulsatile signals. Physiol Meas 2012;33:1491–1501. doi: 10.1088/0967-3334/33/9/1491.
83. Shcherbina A, Mattsson CM, Waggott D, Salisbury H, Christle JW, Hastie T, et al. Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. J Pers Med 2017;7:E3. doi: 10.3390/jpm7020003.
84. Vetrovsky T, Siranec M, Marencakova J, Tufano JJ, Capek V, Bunc V, et al. Validity of six consumer-level activity monitors for measuring steps in patients with chronic heart failure. PLoS One 2019;14:e0222569. doi: 10.1371/journal.pone.0222569.
85. Herkert C, Kraal JJ, van Loon EMA, van Hooff M, Kemps HMC. Usefulness of modern activity trackers for monitoring exercise behavior in chronic cardiac patients: validation study. JMIR Mhealth Uhealth 2019;7:e15045. doi: 10.2196/15045.
86. Coravos A, Doerr M, Goldsack J, Manta C, Shervey M, Woods B, et al. Modernizing and designing evaluation frameworks for connected sensor technologies in medicine. NPJ Digit Med 2020;3:37. doi: 10.1038/s41746-020-0237-3.
87. Wang C, Zhao SY. Research on an improved incremental updated algorithm for Parallel Association Rule. Comput Technol Devel 2018;28:48–52. Available from: https://kns.cnki.net/kcms/detail/61.1450.TP.20180307.1431.070.html.
88. Hasselgren A, Kralevska K, Gligoroski D, Pedersen SA, Faxvaag A. Blockchain in healthcare and health sciences-a scoping review. Int J Med Inform 2020;134:104040. doi: 10.1016/j.ijmedinf.2019. 104040.
89. Xu K, Lu Y, Takei K. Multifunctional skin-inspired flexible sensor systems for wearable electronics. Adv Mater Technol 2019;4:1800628. doi: 10.1002/admt.201800628.
90. Harb A. Energy harvesting: state-of-the-art. Renew Energy 2011;36:2641–2654. doi: 10.1016/j.renene.2010.06.014.
Keywords:

Wearable technology; Cardiovascular healthcare; Continuous monitoring

Supplemental Digital Content

Copyright © 2022 The Chinese Medical Association, produced by Wolters Kluwer, Inc. under the CC-BY-NC-ND license.