Using Data From Fitness Trackers and Health Apps to Help People Exercise More
Early in her life, Ipek Ensari discovered the connection between mood and movement. Whenever she became anxious as a child, her mother would suggest they stroll around the block to “walk it off.” And it worked. Those walks made her wonder why she felt considerably calmer and less anxious afterward. As she got older, she realized many people do the same thing–they use physical activity to relieve anxiety or to think more clearly.
“Using something so accessible – physical activity – to regulate one’s mood and anxiety and improve health became a main interest of mine later when I studied psychology in college,” she says.
As a postdoctoral research fellow at the Data Science Institute, Ensari hopes her research will help people exercise more and become healthier. For her project, she’ll analyze data from exercise trackers such as Fitbit and mobile health applications to develop personalized exercise regimens and behavior modifications that improve people’s fitness. Since physical inactivity is the fourth-leading risk factor for death, her research can be a step toward improving the nation’s public health.
Millions of people now wear fitness devices or use health apps to track their physical activity, heart rate, blood pressure, caloric intake, mood and other health behaviors – which can provide a goldmine of digital health data, Ensari says. She intends to use the data to chart out a new field: exercise data science. Combining biobehavioral exercise science with data science, she will try to find the best way to optimize a person’s exercise regimen and improve his or her health. One of her goals is to create a personal “DNA” from individual data and use it to prescribe an exercise regime most likely to improve a person’s health. Ideally, Ensari hopes to create interfaces that will deliver prompts to participants through their wearable devices or mobile apps at times when their data predict that they’d be most receptive to messages.
She intends to design a year-long study looking at digital health data from several dozen participants. If each participant provides 10-15 health indicators a day, her data set will be in the hundreds of thousands. She hopes to predict activity patterns in the data and design personalized exercise regimens and behavior-modification plans for the participants. If the data tell her, for example, that someone’s energy level is high in the morning, she might send the person a message suggesting that he or she exercise early. If the data tell her that jogging is better than cycling for a person’s pain and mood management, she’d advise him or her to jog more and cycle less. She anticipates she’ll be able to also tell the person exactly at what “dose” to jog (i.e., intensity, frequency, duration).
Ensari will work at DSI under the direction of Professors Noemie Elhadad and Suzanne Bakken. Elhadad is an expert in natural language processing and data mining, and Bakken focuses on biomedical informatics and data visualization.
Ensari came to the U.S. from Turkey for her undergraduate work in psychology at Tufts, attracted by “the breadth of research that’s carried out in the U.S.”
“I also came because of the diversity of the students at universities in the U.S.,” she adds. “Being in an environment that promotes and prioritizes diversity was and still is a main driver of my choices for picking where I went to school, and it’s also why I chose to come to Columbia after I completed my Ph.D.”
Ensari’s interest in psychology comes from “being a lifelong shy introvert,” she concedes. “I was always the observer in the corner – the ‘wallflower’ – so had an inherent desire to understand human behavior.” At Tufts, she began formally studying the relation between exercise and anxiety, observing that those who exercised regularly seemed to respond with less anxiety to stressful situations.
She continued studying the anxiety-exercise relationship at Columbia’s Teachers College, where she earned a master’s degree, after which she did a doctoral program in bio behavioral exercise science at the University of Illinois. Her dissertation focused on the effects of yoga on anxiety and panic in women. She herself had become a practitioner of yoga, which she finds immensely calming.
“I am at my best when I am moving and my dissertation was partially inspired by my intensive, regular yoga practice at the time,” she says, “which by the way is among the most commonly reported alternative methods for managing health symptoms such as pain and anxiety.”
It was during her doctoral studies that she also had a realization: Since not everyone responds similarly to a given workout – 30 people cycling at moderate intensity will respond with different physiological and psychological symptoms – to maximize health benefits exercise regimens must be personalized. So as part of her postdoctoral project, she’ll craft proper workout “doses” for each participant.
Ensari is excited to start her fellowship. Her goal is to become a leading investigator in the emerging field of exercise data science. As it is, millions of people use fitness bands and smartphone apps, she says, but the ocean of data generated daily from the devices is not being used to its full potential. Whereas companies design and sell fitness devices for profit, as an academic Ensari intends to use the data generated by these mobile devices to help people improve their fitness and health.
“We have very little research evidence that these health gadgets are doing what they are being promoted as doing,” she says. “Yet many people are already addicted to their mobile devices, so why not use all the data they already provide to help them improve their health? My research is an example of using digital health data for good, not for profit. In this way, my research is very much in line with the DSI’s mission – to use data for good.”
By Robert Florida