Counting steps via smartphones leads to clues about obesity trends

Stanford researchers collected motion data from smartphones as a way to measure activity across hundreds of thousands of people to help figure out why obesity is a bigger problem in some countries than others.

- By Tom Abate

Using step data captured by smartphones, researchers have defined a new public health risk they call “activity inequality.” This occurs when large gaps develop inside a country between people who walk a lot and those who walk very little, leading to unhealthy levels of obesity.
Tim Althoff

Stanford researchers using smartphones to track the activity levels of hundreds of thousands of people around the globe made an intriguing discovery: In countries with little obesity, people mostly walked a similar amount per day. But in countries with higher levels of obesity, there was a big gap between people who walked a lot and those who walked very little.

The researchers used data captured from smartphones to analyze the physical activity of 717,527 men and women from 111 countries, whose steps were studied for an average of 95 days. A paper describing the findings was published online July 10 in Nature. The lead author is graduate student Tim Althoff. The senior author is Jure Leskovec, PhD, associate professor of computer science.

“If you think about some people in a country as ‘activity rich’ and others as ‘activity poor,’ the size of the gap between them is a strong indicator of obesity levels in that society,” said study co-author Scott Delp, PhD, professor of bioengineering and of mechanical engineering.

The researchers dubbed this phenomenon “activity inequality” to evoke the well-established concept of income inequality.

A related finding was the powerful role that gender played in country-to-country differences. Prior studies of physical activity, done mainly in the United States, have shown that men walk more than women, and this was borne out in the global findings. What surprised researchers, however, was how greatly this gender step gap varied from country to country, with negative consequences for women.

“When activity inequality is greatest, women’s activity is reduced much more dramatically than men’s activity, and thus the negative connections to obesity can affect women more greatly,” Leskovec said.

The researchers, who are sharing their findings on an activity inequality website, hope their work will help improve public health campaigns against obesity, and support policies to make cities more “walkable.”

Smartphones and steps

Smartphones are equipped with tiny sensors called accelerometers that can automatically record stepping motions. The researchers acquired the data for this study from the Azumio Argus app, which tracks physical activity and other health behaviors. Azumio de-identified the data but provided key health demographics: age, gender, height and weight. The last two data points enabled the researchers to calculate each person’s body mass index.

The findings leaned most heavily on data from the 46 countries for which Azumio provided at least 1,000 de-identified users — enough to form the basis for statistically valid inferences. The analysis disclosed strong correlations among activity inequality, the gender-activity gap and obesity levels.

“For instance, Sweden had one of the smallest gaps between activity rich and activity poor, and the smallest disparity between male and female steps,” Althoff said. “It also had one of the lowest rates of obesity.”

Meanwhile, the United States ranked fourth from the bottom in overall activity inequality, indicating a large gap between activity rich and activity poor. It was fifth from the bottom in the gender step gap, and it has high levels of obesity.

Walkable cities

To better understand the causes and consequences of activity inequality in urban settings, the researchers analyzed a large subset of data from the United States to investigate how the built environments of 69 cities related to activity, obesity and health.

In cities that are more walkable, everyone tends to take more daily steps.

Prior research had scored each city by how walkable and pedestrian-friendly it is, using factors such as ease of walking to shops, restaurants, parks and other destinations. The researchers then correlated this walkability index to the smartphone activity data they had collected.

Co-author Jennifer Hicks, director of data science for the Mobilize Center at Stanford, said the results make clear that city design has health impacts: The cities that were most conducive to walking had the lowest activity inequality.

“Looking at three California cities in close geographic proximity — San Francisco, San Jose and Fremont — we determined that San Francisco had both the highest walkability score and the lowest level of activity inequality,” she said. “In cities that are more walkable, everyone tends to take more daily steps, whether male or female, young or old, healthy weight or obese.”

A new research instrument?

The technological star of the project was the increasingly ubiquitous smartphone. Nearly 70 percent of adults in developed countries now carry smartphones; in developing nations, the percentage is close to half.

“This opens the door to new ways of doing science at a much larger scale,” said Delp, who is also director of the Mobilize Center and the James H. Clark Professor in the School of Engineering.

But qualifying the smartphone as a tool for this type of research was no cakewalk.

“The methodology was so new that the reviewers were dubious at first,” Leskovec said.

But strong data and rigorous computational methods ultimately proved the validity of this new approach. Now, having qualified the smartphone for research of this sort, the Stanford researchers are looking for new ways to leverage this tool.

“With the appropriate apps and sensors we can push this research in exciting directions,” said co-author Abby King, a professor of medicine and of health research and policy. “We could better link activity within and across populations with food intake, or examine the ways activity and inactivity may affect stress or mental health, as well as investigating how best to fine-tune our environments to promote increased activity.”

Rok Sosic, PhD, a senior research engineer in computer science, also co-authored the paper.

The research was supported by the National Institutes of Health, SAP, the National Science Foundation and the Stanford Data Science Initiative.

Glen Martin, a freelance science writer, and Raymond MacDougall, lead communications specialist at the National Institute of Biomedical Imaging and Bioengineering, contributed to this article.

About Stanford Medicine

Stanford Medicine is an integrated academic health system comprising the Stanford School of Medicine and adult and pediatric health care delivery systems. Together, they harness the full potential of biomedicine through collaborative research, education and clinical care for patients. For more information, please visit med.stanford.edu.

2023 ISSUE 3

Exploring ways AI is applied to health care