The Boom in Wearable Technology: Cause for Alarm or Just What is Needed to Better Understand Sleep?

Wearable technology is ubiquitous in western society. Millions of low-cost, high-tech devices have been sold worldwide and consumers are using them to monitor multiple health-related signals, including sleep.^1–3 In clinical settings, patients presenting with various sleep disturbances increasingly show up with data from wearables on their smart phones. Clinicians are often befuddled when asked to interpret these data because there are no accepted guidelines or standards on how to interpret it, in part because the FDA does not regulate consumer-level wearables that provide “general wellness” information.³ Although the majority of the wearable technology industry claim that their products are not intended for scientific or medical purposes, researchers and clinicians focusing on sleep are waking up to the idea that these inexpensive tools could be used to monitor research participants and patients, and have begun using wearables in research and clinical practice. Given the widespread use and popularity of these devices, they could also be a new resource for healthcare providers to track a patient's behavior, such as sleep patterns, in relation to disease progression or treatment over a long period of time. Thus, wearables are not only consumer products but are also part of the health information technology revolution that aims to improve health information management, prevent medical errors and reduce health care costs.

For wearable technology that purports to measure sleep to be accepted by the research and medical community, it needs to be vetted and have proven reliability and accuracy. While several devices on the market have been shown to be reasonably accurate in specific populations and applications,³ there are conflicting results from studies attempting to validate sleep trackers against gold standard polysomnography.³ Some of the discrepancies likely relate to differences in the populations being studied, such as adolescents^4–6 versus adults,^7,8 and mixed groups with and without sleep disorders,^5–7 which is important because, similar to standard actigraphy,⁹ commercial devices perform less well compared to PSG in persons with frequent awakenings and/or periods of immobile wakefulness. Discrepancies also arise through testing different models of devices. For example, studies evaluating different generation models of Fitbit (Fitbit, Inc., San Francisco, CA, USA), i.e. Ultra, Classic, FitbitChargeHR, compared with PSG, showed vastly different discrepancies when compared to PSG.^5,8,10 Adding to the confusion of whether or not commercial devices are valid is the attempt of some companies to measure time spent in “light” or “sound” sleep, or “dream” sleep, with little information provided how these sleep states are defined or measured. There is simply insufficient evidence to derive broad research and/or clinical uses at the present time.³ Clearly, more validation studies are needed both in the laboratory and at home. A more pressing need is for an active dialog between the sleep community and wearable companies to optimize outputs of these devices so that they can be better applied in research and clinical settings, and more accurately measure sleep quantity and quality for their customers.

There are several significant barriers to understanding the validity of the data collected by consumer-level wearables for the purpose of measuring sleep. These include the lack of availability of technical information such as sensors' accuracy and algorithms used, details on output metrics that are typically considered trade secrets by wearable companies and thus not shared publicly, and access to the raw data beyond the all-night summary. Raw data should be in the form of accelerometer generated vector magnitude (activity) measurements recorded at least at the minute-by-minute level. Another barrier is caused by the sleep community rather than the wearable companies. The sleep community needs to make a collective effort to identify necessary and sufficient requirements for validation research, including agreement on the standard analyses (e.g., Bland-Altman plots, epoch-by-epoch [EBE] analysis) to be performed and quantitative rules to interpret the outcomes. The fast and uncontrolled wearable boom calls for a firm position in setting standards for their use. For example, while there have been attempt to introduce “a priori set clinical cut-offs” for Bland-Altman plot sleep outcomes, sensitivity (ability to detect sleep) and specificity (ability to detect wake), the main outcomes from the EBE analysis, do not yet have standardized rules to conclude if a device performs well or not. Actigraphy-based devices typically have lower specificity than sensitivity,⁹ but it remains unclear what is an acceptable and meaningful level. In addition, EBE analysis on a min-by-min basis is still largely used and accepted instead of a 30-s comparison that would be more in line with PSG sleep staging. A sleep task-force could be created to help set standard guidelines and criteria in the use of wearable technologies in sleep assessment and management. At the same time we need to persevere in the attempt to increase collaborations between the sleep community and wearable technology companies, such as the National Sleep Foundation's initiative that launched a sleep technology council to support and facilitate innovation in consumer sleep technology in 2015. Acceptance and adoption of standard metrics for validation of the many commercial devices available would inform not only research but also companies interested in research applications for their wearable devices. It would also lead to a faster turnaround of validation studies, which is critical in the current environment in which the wide expansion and availability of sleep trackers clearly outpaces the processes involved in validating these devices. The paradox is that information about accuracy and reliability of a specific device becomes available when that particular model is no longer produced or when the algorithm used at the time of the validation has been updated.

A promising solution that could allow greater access to data from commercial sleep trackers may be offered by third party platforms. These emerging services already allow researchers to obtain sleep measures, with greater resolution and data control, compared to standard consumer devices dashboards. For example, Fitabase (Small Steps Labs LLC) now offers access to pre-processed Fitbit data on a min-by-min basis; the next step could be direct access to raw data that will allow the application of independent, publically available algorithms, facilitating the standardization of processing methods and sleep outcome measures.

Consumer wearable devices are capable of collecting vital signs together with sleep metrics from millions of individuals 24 hours a day, 365 days a year, offering an unprecedented window on users' health and sleep. Ultimately, a shared vision within the sleep community will enhance the transition of new valid, reliable and user-friendly tools from the consumer space into the scientific world. Joint efforts are needed between the wearable tech industry and the scientific sleep community to facilitate this transition.

CITATION

de Zambotti M, Godino JG, Baker FC, Cheung J, Patrick K, Colrain IM. The boom in wearable technology: cause for alarm or just what is needed to better understand sleep? SLEEP 2016;39(9):1761–1762.

DISCLOSURE STATEMENT

The authors have indicated no financial conflicts of interest.