Their analysis showed the strongest signals of caries clustering in very young children — who also displayed patterns of iron and vitamin D deficiency — and in older adults. (iStock)
Is there a link between pollutant exposure and dental caries risk? Using artificial intelligence to analyze large U.S. health datasets, researchers at the University of Pennsylvania say they have uncovered previously unrecognized patterns that broaden how dental caries risk may be understood.
Researchers at Penn Dental Medicine analyzed data from the National Health and Nutrition Examination Survey using machine-learning methods and found novel associations linking dental caries with environmental exposures, nutritional indicators and systemic health markers.
The December 2025 study, published in the Journal of Dental Research, identified associations between caries status and factors such as lead and other pollutant exposure, specific laboratory markers, food types and sleep patterns — particularly among children under five and adults aged 65 and older.
The research was led by orthodontics professor Hyun (Michel) Koo and biostatistics and epidemiology professor Jason Moore.
“This kind of machine-learning pipeline can turn complex national health data into clearer hypotheses and better predictive models — starting with oral health, and potentially extending to other areas of medicine,” Koo said in a Penn Dental Medicine news release. He is co-founding director of the university’s Center for Innovation & Precision Dentistry.
Related: Why most dental practices are still on the fence about AI
Cleaning complex national data
NHANES, overseen by the Centers for Disease Control and Prevention, has collected nationally representative health and nutrition data in two-year cycles since 1999 through interviews, physical examinations and laboratory testing. While the survey is considered one of the most comprehensive population-level health datasets in the United States, its size and non-uniform structure present challenges for advanced analysis.
To address this, the researchers developed a data-cleaning pipeline that incorporates a novel outlier-detection algorithm and unsupervised machine learning. The approach allowed them to identify distinct caries-related subtypes and age-driven heterogeneity within the NHANES dental data.
Their analysis showed the strongest signals of caries clustering in very young children — who also displayed patterns of iron and vitamin D deficiency — and in older adults.
“Our results point to the importance of age-targeted prevention and prediction — especially for young children and older adults — guided by real-world diet patterns, laboratory signals and environmental risk context,” Koo said.
Related: Harvard dental researchers explore AI for earlier head and neck cancer detection
High lead levels
The study also found that while individuals with caries had higher blood lead levels — consistent with earlier research — they also showed elevated levels of other markers, including cadmium and cotinine, suggesting that lead exposure may reflect broader environmental risk rather than a single causal pathway.
Nick Jakubovics, editor-in-chief of the Journal of Dental Research, said the findings highlight the need for more precise, group-specific approaches to caries prevention.
“The next challenge is to build on this information and find more effective methods to prevent caries in different groups of people,” he said.
The authors noted that the current analysis is limited to NHANES 2017–2018 data and said future multi-cycle studies will be needed to assess trends over time.
You can listen to our podcast on: AI trends to watch for in 2026 | Episode 35