Smartphone data shows the link between fast food restaurants and diet-related diseases

How many fast food restaurants do you come across every day and what does that have to do with your health? A lot ofsays Abigail Horn, chief scientist at USC’s Information Sciences Institute (ISI).

Horn led a multidisciplinary team with researchers from three USC schools (Viterbi School of Engineering; Dornsife College of Letters, Arts and Sciences; and Keck School of Medicine), MIT and Sabancı University in Turkey; and worked with the LA County Department of Public Health. They wanted to explore whether smartphone mobility data (i.e. location data) could provide a way to measure people’s individually experienced dynamic food environments, at scale across large and diverse populations and diverse physical environments.

The question was: can we use mobility data to measure people’s visits to food stores? Because that’s a good benchmark for eating food at that outlet. And can we then go a step further to see whether visits to food stores, observed in the mobility data, are predictive of the number of nutritional diseases among people?”

Abigail Horn, chief scientist at USC’s Information Sciences Institute

Location, location, location

“It is well known that the physical environment can influence people’s eating decisions and therefore their diet-related health outcomes, but what we don’t know is to what extent that is true,” said Horn, a research assistant professor at the University of California. Daniel J. Epstein Department of Industrial and Systems Engineering at USC Viterbi School of Engineering.

Physical food environments are the actual spaces where people acquire food. “The food stalls in their neighborhood, or around their workplace, or any location on their daily path. Things like supermarkets, restaurants or neighborhood markets,” Horn explains.

These environments have been shown to impact people’s diets and therefore health outcomes in different ways – including diet-related diseases. First, Horn said, “If people have little physical access to healthy foods, that can lead to unhealthy choices out of convenience or necessity.” And second, “People can be aroused by the food environment. So, for example, seeing fast food restaurants over and over again throughout the day can encourage or trigger certain behaviors” (i.e., eating more fast food).

There are a number of studies that look at the food environment in people’s homes and associate it with food choices and diet-related diseases. But the findings are mixed, as are the results of public health initiatives that have focused on the home neighborhood food environment.

Horn explained: “Over the past decade, more than a billion dollars have been invested in public health interventions in the home food environment. This could mean building a supermarket in a food desert. [a home neighborhood with limited access to nutritious food] or stocking the corner stores in that neighborhood with fresh fruits and vegetables.” But she continued, “There has been no measurable impact on increasing healthy food purchases or people’s health outcomes. So what’s going on here?”

Kayla de la Haye is one of the members of the research team who could help answer that question. De la Haye is director of the Institute for Food System Equity at the USC Dornsife Center for Economic Research and has a background in public health, nutrition and psychology. “One of my roles in this research was to bring expertise in how people make decisions about what to eat, and the consequences of food environments that flood people with unhealthy options and put them at risk for many diet-related diseases such as obesity and diabetes .”

Looking beyond the neighborhood market

De la Haye has worked with families across LA – from Lancaster to the East Side of LA – helping them with strategies to avoid unhealthy foods and adopt healthier eating habits. She said, “So I brought this real-world knowledge of the challenges Angelenos face in eating a healthy diet to our research project.”

The team knew from their own experiences and from the experiences of families they have worked with in healthy eating programs that people don’t just eat in their own neighborhoods. But they needed the data to prove this on a population scale. Horn said, “We thought the lack of data on all the places people actually go to eat and where they spend the most time might explain why we don’t see a link between the food environment in the home neighborhood and people’s diet health outcomes.”

So they turned to smartphones for the data.

For most of us, our smartphone is always tracking our location, and we probably share that data with various apps. Location data companies collect this data – called “mobility data” – and sell it for advertising. But it is increasingly being made available for research, for example by Spectus.ai through their Social Impact Program, which provided the data for this study.

Esteban Moro led the team at MIT that would help unlock and analyze this data. Moro, a research scientist at MIT Connection Science, said: “Our group has extensive experience analyzing and using mobility data for problems such as segregation, transportation, urban planning and commercial operations. We are experts in analyzing large data sets of human behavior and by turning them into insightful tools for urban problems. So our main role in this research was to provide and analyze population-wide mobility data on food consumption.”

Bringing all the data together

Using census block data for Los Angeles County to indicate residential neighborhoods, and large mobility data to track daily journeys, the researchers were able to see all the proximity – the “exposures” – that people would have to food stores throughout their days.

The team looked specifically at fast food restaurants, because fast food is commonly consumed and strongly linked to disease risks. Using point of interest data, they identified fast food restaurants in LA County. To bring out the health piece of the puzzle, they accessed survey data from the LA County Health Department.

“The Los Angeles County Health Department conducts a health survey of the LA population every three years. We partnered with them and they were able to share with us anonymized individual-level data on socio-demographics, obesity rates, diabetes rates and very important is the frequency of fast food intake for a representative sample of the LA population,” Horn said.

By analyzing the data, the researchers confirmed that your home neighborhood matters when it comes to your risk of diet-related diseases, but so does your commute, the path you take to get your daily groceries and how you get from point to point. A comes to point B. and all the way to point Z in your day, and what those points are.

The results?

“We know there is a correlation between visits to fast food restaurants and fast food intake, and between fast food intake and diet-related diseases, but wow, this data source does that really well!” said Horn.

Moro explains: ‘The most surprising result is that mobility data acts as a ‘fair signal’, that is, visits to fast food restaurants are a better predictor of individuals’ obesity and diabetes than their self-reported fast food intake, controlling for other known risks .”

De la Haye highlighted: “This work shows that large-scale mobility data is in fact a valuable indicator of where and what people eat, and of their risk of diet-related diseases.”

Why is this so important?

De la Haye explains: “Measuring what people eat is very difficult. In fact, many major public health surveys and surveillance tools have stopped asking people about their food intake because the data is often unreliable (in part because people often forget the details of what they eat). they ate, and also because they don’t always want to tell researchers about their less healthy food choices. So this gives us a new tool to track dietary patterns, such as fast food eating, for large populations such as residents of cities, counties, or the entire country.”

What’s next?

‘What I’m excited about as a researcher,’ says Horn, ‘is that this opens up mobility data for all kinds of studies into the food environment. Things like: where do people get food at different times of the day? People? When are they most affected by the options available (or unavailable) to them? We can really explore this with large mobility data because it allows us to look at eating behavior in large and new dimensions: at scale for the entire population, across the entire population, diverse populations, diverse ecological environments, and over long periods of time.”

De la Haye underlines the importance of this: “data on population diets is a powerful tool needed to create public health programs and policies and ultimately reduce health risks from one of the leading causes of disease and death in the US: unhealthy food. “