Forty-one point nine percent of American adults live with obesity. It costs the U.S. healthcare system $173 billion every year. The causes are well-catalogued: poor diet, physical inactivity, socioeconomic inequality, food deserts. But there is one cause that public health researchers have been remarkably slow to study with the rigor it deserves — the recommendation algorithm.
Every day, social media platforms in the United States serve an estimated 4.6 billion algorithmic content recommendations. A significant proportion of those recommendations are for food and beverage products. And the evidence is accumulating that the foods being algorithmically amplified are disproportionately the ones fueling the obesity epidemic.
"The digital environment shapes behavior as powerfully as the physical environment. We regulate what ends up on grocery store shelves. We are only beginning to ask who regulates what ends up in the social media feed."
What My Research Is Examining
My first flagship research paper in the health analytics area introduces the Social Media Obesity Risk Index (SMORI) — a composite analytical framework that quantifies the relationship between algorithmically delivered food advertising intensity and regional obesity prevalence across U.S. states and metropolitan areas.
The SMORI framework draws on publicly available data from the Meta Ad Library API, Google Ads Transparency Center, and CDC's Behavioral Risk Factor Surveillance System (BRFSS). By combining these sources, we can begin to answer questions that have never been rigorously examined at scale:
- Are regions where social media users receive more unhealthy food advertising also regions with higher obesity rates?
- Do lower-income communities receive disproportionately higher volumes of algorithmically targeted unhealthy food advertising?
- Can machine learning models trained on ad transparency data generate predictive regional obesity risk scores?
Why Marketing Analytics — Not Just Epidemiology — Matters Here
Epidemiologists are excellent at documenting disease patterns. But understanding the commercial data infrastructure that shapes those patterns requires a different toolkit — one that combines advertising analytics, platform algorithm mechanics, consumer behavior modeling, and geographic information systems. That intersection is precisely where marketing analytics researchers should be working.
The platform recommendation algorithm is, at its core, a marketing analytics system. It is optimized for engagement and commercial conversion. If we want to understand how it shapes public health outcomes, we need researchers who understand both the technical architecture of those systems and the population-level health consequences they produce.
The Policy Connection
This research is directly relevant to active federal policy proceedings. The FTC's commercial surveillance rulemaking is examining exactly this question: what obligations do platforms have when their advertising systems produce predictable harms? The White House AI Executive Orders explicitly identify algorithmic accountability as a national priority.
Research in Progress
The SMORI framework paper is currently in preparation for submission to the Journal of Marketing (AMA) and American Journal of Public Health. The study uses a panel regression approach with state-level obesity rates as the dependent variable and platform-level algorithmic food ad intensity as the key independent variable, controlling for income, education, race/ethnicity, food desert status, and physical inactivity.
What You Can Do
If you are a public health researcher, a marketing analytics scholar, or a policy advocate who sees the connection I am drawing here, I would welcome your perspective. The interdisciplinary community working at the intersection of marketing analytics and public health is still small. It needs to grow — and grow quickly.
The algorithm is already shaping the waistline of America. The question is whether research will catch up fast enough to do something about it.