In recent years, wearable sensors, such as FitbitⓇ and Apple WatchⓇ, have become increasingly popular, allowing users to track their physical activity, sleep, heart rate, and other health-related metrics. But in addition to telling you how many calories you burned on your last jog, these devices also have immense potential for scientific research, as they can provide rich data about human behavior, physiology, and health in real-world settings. Researchers at the University of Arizona Health Sciences are among many who seek to use wearable sensor data to shepherd in the next generation of scientific breakthroughs.
Collecting and analyzing sensor data from wearables can be expensive and time-consuming, requiring specialized expertise in data science, statistics, and programming. That's where UArizona’s Center for Biomedical Informatics and Biostatistics (CB2) comes in. Shravan Aras, PhD, is the Assistant Director of Sensor Analysis & Smart Health Platforms (SASP) at CB2. His team seeks to bridge the gap between the mass of raw data from wearable sensors and the ability to make that data scientifically useful.
The SASP effort began in April of 2022. Dr. Aras’s first step was to interview faculty, staff and students in Nursing, Physiology, and Nutrition Science at UArizona, which uncovered several unmet needs around integrating sensor data into studies. So the SASP team set out to develop a solution that addressed them all. Today, SASP supports researchers from the study planning phase through project closing, including support for grant writing, sensor integration, data analysis and visualization. Central to this work is SensorFabric–a searchable, centralized repository for sensor data. “The name ‘SensorFabric’ came from the theory of relativity,” says Dr. Aras, referring to the “fabric of spacetime,” in which time and space are woven together. In the same way, his vision is that SensorFabric’s elements will be woven into a cohesive, user-friendly whole, enabling searchability and collaboration throughout the university.
Until now, UAHS has had no standard methods for collecting, storing and searching sensor data. SensorFabric elegantly addresses this issue by providing a single hub for all of these functions. Dr. Aras explained that data comes in from different devices in a wide variety of formats. SensorFabric funnels the data through various tools and programs, converting it all into a single, standardized format compatible with common tools like Power BI, Tableau, PyTorch, RStudio and more. This allows for quick and easy data analysis and visualization–and investigators can view data from multiple sensor types on one centralized dashboard. SensorFabric not only stores the sensor data itself, but any participant surveys associated with it (The previous process for merging sensor data and surveys took three to four months of coding and data entry; SensorFabric automatically links the two right from the beginning.)
A researcher’s main tool for engaging with their participants is usually a smartphone app–and SASP provides this service as well. Mobile apps are as unique as the studies they support, each requiring a customized interface. But instead of having to build each application from the ground up, SASP offers an easy-to-use “low-code/no-code” option. CB2 has partnered with Care Evolution, a company specializing in health and research-related data products. “Care Evolution has been a great partner,” said Dr. Aras. Their product, MyDataHelpsTM, allows researchers to create their own apps by dragging and dropping components–no coding knowledge required–and is fully secure and HIPAA-compliant. This is a game-changer for researchers and the teams who support them. “To put it in perspective… [app development] used to take us almost six to seven months with a team of five people programming. MyDataHelpsTM allows us to do everything in a matter of weeks,” says Dr. Aras. MyDataHelpsTM is compatible with all major sensor brands, and even some geographic and weather databases.
Still, one of the biggest challenges with integrating wearables into a study is compliance–study participants often forget to charge their device, damage it, or simply lose interest, creating pockets of missing data. This can harm the quality of both the data and any research findings it produces. In the past, compliance could only be measured at the end of a study, which was too late to intervene. SensorFabric has a compliance platform that gives researchers daily updates throughout the study, flagging users who aren’t wearing or syncing their devices consistently. This way, researchers can send reminders to gently nudge participants as needed. The SASP group is also working to leverage Amazon Alexa and other “smart platforms” to improve compliance. CB2 sponsored a UA Capstone team project called Antaris that enables participants to engage with studies via voice. For example, a participant in a nutrition study could simply ask Alexa to record their meals for the day. This solution dovetails with MyDataHelpsTM, improving accessibility and ease of use.
SensorFabric can also query data on an hourly basis, enabling “just-in-time” interventions–for instance, if someone doesn’t meet their step goal for the day, they can get a notification through their Alexa device encouraging them to be more active tomorrow. In the past, researchers couldn’t analyze sensor data until days later–often too late to effectively respond. “It would take many weeks in a traditional setting for somebody to integrate FitbitⓇ into a study and get all the data. But if someone comes to us today, and tells us, ‘hey, I have a new study,’ it would take us about a week or two just to build all the surveys and consent forms… and then once the study goes live, it’s only a matter of a few days.” said Dr. Aras. The centralized nature of SensorFabric also creates accountability and traceability of data, which are requirements for many publications and grants. Even with all of this capability, Dr. Aras says that SensorFabric is “still in its infant state,” and has the capacity to grow into an indispensable tool for the UAHS research community.
Wearable sensors are just starting to emerge as the future of real-time health research and intervention, with CB2’s Sensor Analysis & Smart Health Platforms at the cutting edge. To date, SASP has supported studies in athletics, nutritional sciences, and diabetes research, and has plans to expand their services to a pipeline of future projects. By providing a user-friendly platform for every facet of sensor data integration, SASP helps UAHS researchers leverage the power of wearables to deepen our understanding of human health.
Sources:
Abdullah, K., Chiu, W., Mac, S., Sanchez, D., Setiadi, J., & Messenger, R. (2023). Antaris - An IoT based system for at-home health interventions [Photograph: Antaris informational poster]. University of Arizona Health Sciences Center for Biomedical Informatics and Biostatistics (CB2). https://cb2.uahs.arizona.edu/services/sensor-analysis-and-smart-health-p...
Hew, Y., Pal, A., & Aras, S., PhD (2023). SensorFabric: M-Health App Development & Integrating Fitbit, Apple Watch, Google Fit [Photograph: SensorFabric informational poster]. University of Arizona Health Sciences Center for Biomedical Informatics and Biostatistics (CB2). https://cb2.uahs.arizona.edu/services/sensor-analysis-and-smart-health-p...