Foodborne illnesses, such as the recent romaine lettuce outbreak, are increasingly common in the United States, with an estimated 1 in 6 people affected each year. Image/Marco Verch
Abigail Horn is passionate about food.
The new Research Assistant Professor in the Daniel J. Epstein Department of Industrial and Systems Engineering has built her career by using data and computer science to help us understand eating behavior, diet, and health, and to combat worrying increases in foodborne disease outbreaks.
“I’m very food-motivated,” Horn laughed. “The most basic thing after air and water, is food.”
“But I’ve always wanted to have a very clear public health impact. When I started my Ph.D. in engineering systems at MIT, I started researching large-scale outbreaks of foodborne disease — and found that there’s a huge problem in tracing the source of these outbreaks in the supply chain. They’re not so frequent, but because we can’t quickly identify their sources, when they do happen, they cause many illnesses and create a lot of fear,” Horn said.
Abigail Horn is an expert in large-scale outbreaks of foodborne disease and works within the Daniel J. Epstein Department of Industrial and Systems Engineering and Information Sciences Institute.
The Centers for Disease Control and Prevention estimate that 1 in 6 people in the U.S. gets food poisoning each year — causing around 128,000 hospitalizations and 3,000 deaths. Recent years have seen increases in large-scale foodborne illness outbreaks. In 2018, a devastating E. coli outbreak affecting romaine lettuce led to the product’s largest ever recall across multiple states in the USA. The outbreak caused 210 people to become ill across 36 states, some hospitalized with kidney failure. The estimated cost to society of the outbreak and recall was up to $343,000,000.
Horn said that it is particularly difficult to narrow down the source of an outbreak, such as those affecting the production of leafy greens like romaine or spinach. The result is that produce is indiscriminately wiped off the shelves, resulting in significant wastage and public mistrust of the food source.
“And we don’t want to see fewer people eating their salad,” Horn said. “I wanted to develop data-driven and model-based approaches to help solve this problem, because in public health response, there hasn’t been a lot of utilization of novel data resources, in large part due to the time demands on responders.”
Horn developed models of the network structure of the food supply chain, and then built models on top of these to trace and infer the source of outbreaks. Horn said that if an outbreak starts in the network — for instance, at a farm — and propagates through the supply chain, the reported individuals getting sick and their associated contamination sites act as “sensors” within the network that allow the model to observe and make predictions about where the contamination could have originated.
Horn’s research modeling work led to the German government contacting her to assist with their foodborne illness response system at the German Federal Institute for Risk Assessment.
“I was working at a federal-level research institute together with logistics modelers who were also developing models of the network structure of the supply chain. We were working to adapt and implement the models so they could use them in actual outbreak settings,” Horn said.
As a postdoctoral fellow at Germany’s Kühne Logistics University, Horn’s work incorporated transport data to better understand foodborne disease contamination transmission across the German food supply system, from retailers to consumers.
Horn, who is also a Research Lead at the USC Viterbi Information Sciences Institute, joins the Epstein Department after her recent work as a Research Associate in the Department of Population and Public Health Sciences at Keck School of Medicine of USC, where her research covered infectious disease modeling, including COVID-19 modeling for Los Angeles County.
What Smartphones Can Tell us About our Food Habits
Most recently, Horn’s work has focused on using large-scale data on human mobility from smartphones to study eating behaviors, drawing on methods from computational social science. Horn said she is interested in how people of diverse backgrounds move around their environments to access food, how this links to their diet and health, and how this informs more optimized strategies for interventions to improve diets. Horn found that many of the data sources that could be harnessed for modeling foodborne disease could also be used to model nutrition.
“Understanding how people have access to food and food environments, the ways that people access food outlets, what food people eat and how that differs across different populations — we do a really poor job of measuring that at the systemic level,” Horn said. “Yet we have all of these data points that are being generated from our digital traces — like mobility, menus, food delivery, retail purchases — that we can study to understand what people are eating, how it differs across populations, and how those disparities perpetuate into public health problems.”
Horn said that most public health policy in the last century has focused on telling people what to eat in order to encourage balanced diets. It’s a policy approach that hasn’t worked, leading to increased efforts to understand the systemic issues that influence diet.
“You may have heard of the concept of the food desert, which is a neighborhood that has low access to fresh food, for example, fresh fruit and vegetables, or what in public health nutrition is called a food swamp, which is a neighborhood where there’s a high saturation of fast-food outlets,” Horn said.
“What I wanted to do with the mobility data is to look beyond the home neighborhood to all the places that people go to throughout the day,” Horn said. “For example, if you work or do errands in a neighborhood that has a certain composition of healthy or unhealthy food outlets, how does that influence you in the food decisions that you make?”
Horn said she was looking forward to collaborating with her new colleagues in the Epstein Department, especially given their substantial optimization and logistics expertise.
“In the Information Sciences Institute, I’m working with really great machine learning scientists and people with natural language expertise and using that to develop aspects of my work, and I’m really excited about merging that with the insight and expertise in ISE around designing optimized policy level interventions. I’m also looking forward to further developing logistics models as they apply to the food supply chain and exploring intersections between large-scale food supply and distribution systems and the accessibility of nutritious, affordable foods in local food distribution systems.”
Published on October 13th, 2022
Last updated on May 16th, 2024
