Smartphone location data shows promise for public health research

A powerful tool for public health research could be in your pocket.

Researchers at Washington State University have pioneered an innovative method for using Google Location History (GLH) data to conduct large-scale longitudinal studies on human behavior and health on topics including outdoor exercise habits, visits to fast food outlets and exposure to environmental pollution.

The method, highlighted in a series of studies, offers a glimpse of a future where "citizen science" based on smartphone data could reshape public health research. Data shared by smartphone users offers a trove of information about exercise habits and other behaviors-and gathering it is far more efficient than older methods such as placing GPS trackers on study participants. Such data could transform population-level studies, and it could eventually result in personalized health reports for phone users, such as a notification of pollution exposure.

Google Location History (GLH) data, an opt-in form of precise location tracking available through Google Maps, is one form of user-generated data with broad applications for public health research and policymaking.

"This data is like nothing else out there because it covers a long period of time in a very granular way," said WSU Associate Professor Ofer Amram, one of the lead authors on the research. "It can give us insight into how human behavior interacts with the environment, and when we link it with health data, we can look at how behavior and environment impact health outcomes."

In a series of studies, collaborators from WSU's Department of Nutrition and Exercise Physiology and School of Electrical Engineering and Computer Science tested the data's usefulness by recruiting 270 members of Washington State Twins Registry who had participated in previous exercise research and were willing to share their GLH data. Participants collectively shared nearly 18 million months of data, with some records spanning over a decade.

The researchers then compared previously collected GPS and accelerometer data with how Google's proprietary algorithm classified movement into activities such as walking, biking, and driving. The resulting analysis, published in the Journal of Physical Activity and Health, found the algorithm accurately classified walking but was not as reliable for running or biking. The study also found a strong association between walking and healthy weight, confirming GLH data produces the same results as other research methods.

The potential applications of GLH data are broad. In a second analysis published in the International Journal of Health Geographics and led by PhD student Olufunso Oje, the researchers found GLH data is detailed enough to study how often people visit fast food outlets compared to other restaurants and grocery stores. This could allow researchers to study how an individual's "food environment," including the physical availability of different types of food, impacts food choices and health outcomes.

"Imagine the difficulty of asking someone to recall where they ate a few years ago. Yet with this data, we can trace detailed food environment exposures over time that would otherwise be impossible to study," Oje said.

GLH data can also be linked to other types of data to provide a more complete picture of population health, such as data on exposure to environmental pollution, a third analysis published in Environmental Health Perspectives found. While most studies use participants' home addresses to estimate their exposure, GLH data enables researchers to estimate exposure wherever participants take their phones.

Additionally, the challenges of working with such enormous quantities of data sparked innovations in computer science, noted Associate Professor Assefaw Gebremedhin, Oje's advisor. The researchers developed a framework for efficiently processing GLH datasets that was published in ACM Transactions on Spatial Algorithms and Systems.

The researchers note that GLH data may raise privacy concerns regarding location tracking. In addition to Google's privacy-protecting measures, researchers can manage this through anonymizing datasets.

This research was supported by funding from the National Institutes of Health, the National Institute of Environmental Health Sciences, and the National Institute of Aging.