Mapping Diet Decisions: How Environment Impacts Food Choices in America

In 2008, Los Angeles passed a law to fight obesity in low-income neighborhoods.
The new law affected some 700,000 residents, restricting fast-food restaurants from opening in certain neighborhoods and prohibiting existing ones from expanding.
“I believe this is a victory for the people of South and Southeast Los Angeles, for them to have greater food options,” Councilwoman Jan Perry said at the time, according to the Los Angeles Times.
But according to a 2015 study from the RAND Corporation, obesity rates actually grew faster in the areas with the fast-food restriction.
Now, Abigail Horn of the USC Viterbi School of Engineering is combining the lessons of the past with new AI insights and novel data streams to help Americans make healthier food choices.
Horn, a lead scientist at the USC Information Sciences Institute and research assistant professor in the Daniel J. Epstein Department of Industrial and Systems Engineering, is currently leading a research project focused on using large-scale human mobility data collected from smartphones to help increase equitable access to healthy food in the U.S. By studying how people’s daily routines and the impact that has on their eating habits, Horn hopes to design better interventions that meet people where and when they need it.
For example, in response to the 2008 law, Horn noted: “When people are going to fast food outlets, our research shows they’re doing that on average five miles away from their homes. So maybe we shouldn’t be focusing on people’s immediate home neighborhood. We should be focusing on the neighborhoods where they are during their day, nearby their workplace, where they run errands, or where they drop their kids off for school, these kinds of things.”
She is currently using the smartphone location data of over 15 million Americans and analyzing it with machine learning techniques to better understand how we make decisions about where and what to eat.
“It’s highly sampled spatial, temporal data,” Horn said. “We must be able to set up the data in a way that we can design ‘natural experiments’ to develop causal inferences from observational data. It’s not like the data was designed for this use.”
Using artificial intelligence, specifically machine learning, Horn aims to draw conclusions from patterns within the data. If there are any elements of the environment (independent of personal preferences) that influence people to visit a less healthy food outlet when they have multiple options, she wants to uncover them. This is what is meant by “causal inferences.”
“There’s this whole field of causal inference and machine learning developed to handle the high dimensionality of complex observational data,” Horn said. “We have to develop new approaches to be able to do that in space and time, respecting our domain of trying to understand how people are affected by their environment and how that impacts their health.”
Although analyzing this data is exceptionally challenging, its scale and the potential to significantly enhance the food environment on a broader level led the National Science Foundation to award Horn $1.5 million to support her research on the project, titled “Smart and Connected Communities Food Environment Dynamics (SCCFED): Leveraging AI and Human Mobility Data to Address Disparities in Food Access and Improve Americans’ Diets.” Her co-investigators on this project are Kayla de la Haye, director of the Institute for Food System Equity and associate research professor of psychology and spatial sciences at USC, and Esteban Moro, professor of physics at the Northeastern University Network Science Institute.
She plans to use part of the money to build a robust research team, including supporting students, as well as to acquire necessary data resources.
Additionally, Horn wants to conduct customer surveys outside fast food outlets to deepen her understanding of consumer behavior called “food outlet intercept surveys,” where you “stand outside a food outlet and ask people not just what they got, but why, and try to understand what induced certain kinds of eating decisions.”
Michelle Wood, program manager of food policy and procurement at the LA County Department of Public Health will partner with Horn’s team on these surveys..
Surveys and direct engagement, Horn added, can help researchers understand how convenience, familiarity and affordability drive decisions, especially in neighborhoods with fewer healthy food options.
Horn is now working with the Los Angeles County Department of Regional Planning to inform the design of policy changes based on some of her team’s research findings. Possible policies could include improved zoning laws to prohibit fast food in certain areas, the zoning of specific areas to incentivize grocery stores where they are needed, or mobile markets or street produce vendors.
On Dec. 19, 2024, Horn held a project launch event for the SCCFED project, with the hope of strengthening the partnership between the research team and the LA County government collaborators to identify priority use cases for research and pilot efforts together.
“Our L.A. County partners are embracing the potential of this large-scale data to proactively evaluate policy changes, trusting that we can draw insights to support these initiatives,” Horn said, “and that’s really exciting.”
Published on January 22nd, 2025
Last updated on January 22nd, 2025