Evaluating the Consistency of Current Mainstream Wearable Devices in Health Monitoring: A Comparison Under Free-Living Conditions

• Published in: Journal of medical Internet research Vol. 19; no. 3; p. e68
• Main Authors: Wen, Dong, Zhang, Xingting, Liu, Xingyu, Lei, Jianbo
• Format: Journal Article
• Language: English
• Published: Canada Gunther Eysenbach MD MPH, Associate Professor 07.03.2017; JMIR Publications
• Subjects: Accuracy; Activities of Daily Living; Adult; Brands; Cell Phone; Cellular telephones; Chronic illnesses; Consumer behavior; Female; Health status; Humans; Investigations; Living conditions; Male; Measurement; Mobile phones; Monitoring, Physiologic - instrumentation; Original Paper; Physical fitness; Product development; Reproducibility of Results; Sensors; Sleep; Social Conditions; Wearable computers; China; health status; monitoring, physiologic; fitness trackers; activities of daily living; motor activity
• ISSN: 1438-8871; 1439-4456; 1438-8871
• DOI: 10.2196/jmir.6874

Wearable devices are gaining increasing market attention; however, the monitoring accuracy and consistency of the devices remains unknown. The purpose of this study was to assess the consistency of the monitoring measurements of the latest wearable devices in the state of normal activities to provide advice to the industry and support to consumers in making purchasing choices. Ten pieces of representative wearable devices (2 smart watches, 4 smart bracelets of Chinese brands or foreign brands, and 4 mobile phone apps) were selected, and 5 subjects were employed to simultaneously use all the devices and the apps. From these devices, intact health monitoring data were acquired for 5 consecutive days and analyzed on the degree of differences and the relationships of the monitoring measurements ​​by the different devices. The daily measurements by the different devices fluctuated greatly, and the coefficient of variation (CV) fluctuated in the range of 2-38% for the number of steps, 5-30% for distance, 19-112% for activity duration, .1-17% for total energy expenditure (EE), 22-100% for activity EE, 2-44% for sleep duration, and 35-117% for deep sleep duration. After integrating the measurement data of 25 days among the devices, the measurements of the number of steps (intraclass correlation coefficient, ICC=.89) and distance (ICC=.84) displayed excellent consistencies, followed by those of activity duration (ICC=.59) and the total EE (ICC=.59) and activity EE (ICC=.57). However, the measurements for sleep duration (ICC=.30) and deep sleep duration (ICC=.27) were poor. For most devices, there was a strong correlation between the number of steps and distance measurements (R >.95), and for some devices, there was a strong correlation between activity duration measurements and EE measurements (R >.7). A strong correlation was observed in the measurements of steps, distance and EE from smart watches and mobile phones of the same brand, Apple or Samsung (r>.88). Although wearable devices are developing rapidly, the current mainstream devices are only reliable in measuring the number of steps and distance, which can be used as health assessment indicators. However, the measurement consistencies of activity duration, EE, sleep quality, and so on, are still inadequate, which require further investigation and improved algorithms.