Is it safe to use consumer wearable devices for research?

Consumer wearables, such as smart watches or rings, can help people to track their daily steps, heart rate and sleep patterns. Given the devices’ popularity and ease of use, researchers are increasingly harnessing them to study health — for example, by investigating exercise regimes or monitoring chronic diseases.

Nature Spotlight: Digital wearables

Wuyoh Sui, a digital-health and behavioural-change researcher at Western University in London, Canada, has been studying how best to use digital technologies to improve peoples’ physical-activity levels and mental well-being. He has also been researching the accuracy, safety and effectiveness of these artificial-intelligence tools used to improve health behaviours, which often rely on ‘black box’ algorithms that make it hard to understand how inputs and outputs are created and handled.

Although wearables will undoubtedly provide more-tailored health interventions by using personalized data, Sui says there are many ethical concerns to navigate when using such commercial devices in research. The biggest two are data privacy (what data are used by the AI-based algorithms in these devices) and data security (who can potentially access those data). Risks include the sale, sharing and theft of sensitive metrics that reveal the daily habits and physical status of a person to advertisers, insurers and companies that sell personal information.

Sui spoke to Nature about how researchers can navigate the complex, shifting landscape of consumer wearables in scientific research.

How did you start to explore the ethical considerations surrounding the use of consumer wearables in research?

Until 2019, my colleagues and I used conventional research-focused wearables such as inclinometers — devices that can be attached to a person’s thigh to determine whether they are sitting or standing. But these devices were expensive and required the wearer to shave their thigh and attach the inclinometer and a waterproof cover.

Will wearable technologies transform clinical trials?

We began to explore ways to collect step counts from the commercial health apps used by study participants. Our aim was to use those data in Pathverse, a digital platform created to deploy custom health interventions to participants. In separate work, we collected data from commercial virtual-reality headsets with built-in accelerometers to corroborate participant’s activity levels. But in both cases, we didn’t have a way to independently assess how these third-party apps determined the frequency and intensity of movement.

Given the lack of transparency around the AI algorithms used by wearable-device companies, platforms that use their data could inadvertently generate faulty conclusions. For example, if the step count reported to the participant is inaccurate, there is a possibility of delivering flawed health recommendations, such as advising a participant to double their step count. This is just one of many ethical concerns.

As my colleagues and I continued to explore a broad range of ethical considerations linked to wearables, we touched on a key question, which we raised in a 2023 paper¹: are we, as scientists, opening up ourselves or our study participants to risks by using consumer wearables?

What are the most pressing ethical issues with using wearable technologies?

At the moment, data privacy and security concerns supercede worries about researchers’ ability to validate non-transparent algorithms. Smartphones and smart watches introduced new opportunities to track movement, but added a third-party developer with commercial interests that are beyond the control of the researcher. As scientists, we are sharing the data with the developer, whether we like it or not, yet we have no control over where those data end up or how they are used. Any entity that collects the data is in a position to broker them.

Let’s say a digital wearable tracks a person’s running habits. At a micro level, the device maker could push specific articles or adverts for products to that person, potentially influencing their behaviour. At a macro level, depending on the company’s terms of service, they could make their user database available to advertisers. These actions could compromise the research process, and raise concerns about security and privacy. Furthermore, device makers can also use the data to train their AI-based algorithms to influence consumer behaviour more effectively.

How have AI concerns informed your work?

In 2021, the United Nations High Commission stressed the need for a temporary pause on the adoption and use of AI systems that pose significant risks to human rights, which made us stop and think about the ethical concerns of using commercial wearables in our research.

If AI becomes conscious: here’s how researchers will know

AI tools enable streamlined interactions between researchers and study participants. For example, when providing exercise advice on the basis of a participant’s current behaviours, researchers could gather information by using an AI chatbot that asks the participant about their step goals or any physical barriers they face. Combined with data from the wearable devices, the answers can be used to create tailored interventions that shape participants’ goals and behaviours in ways that researchers alone couldn’t. Researchers used to create their own large language models (LLMs) to guide chatbot interactions with participants. That’s the better, but harder, way to go. The alternative is to use commercially available systems, such as ChatGPT. To use those effectively, however, researchers need to provide detailed prompts to guide the chatbots’ interactions with study participants — but researchers will have much less understanding and oversight into how those chatbots work and collect data compared with their own LLMs.