Researchers predict cavity risk from social factors in 23,726-person study

Researchers used machine learning to predict cavity risk and severity from social and demographic data covering 23,726 people in Brazil. The findings, published in Caries Research, point toward a way to identify groups needing preventive dental care—not by examining a single tooth, but by examining the conditions surrounding people’s lives.

Wealth and household density were the strongest predictors of cavity risk in the study’s analyses. The researchers also found associations involving dental insurance, recent dental visits and access to fluoridated water. Their proposed use is public-health planning: helping direct resources toward people whose risk can disappear inside community-wide averages.

Learning from households and dental exams

The team analyzed participants in the 2023 Brazilian Oral Health Survey, a national study coordinated by Brazil’s Ministry of Health. It included children, adolescents and adults. Alongside information about their lives and households, participants received clinical exams measuring how common and severe tooth decay was.

That combination let the researchers examine social determinants of health—the circumstances linked to people’s health—alongside measured dental outcomes. The survey captured several kinds of information:

• Demographics, including age, sex and race, allowing the team to examine risk across different population groups.
• Household and socioeconomic conditions, including wealth, household density, private dental insurance status and water source.
• Dental-care behavior, including how much time had passed since a participant’s last dental visit.

The researchers first examined those relationships using traditional analyses, then applied several machine-learning algorithms. The models evaluated the relative contributions of different factors to predict cavity risk and severity by age group. This was an analysis of existing survey records, with dental exams supplying the measured outcomes.

NYU’s account describes the models as accurately predicting those outcomes, but gives no numerical accuracy estimates. Senior author Fernando Neves Hugo says the approach can capture interactions among social factors that traditional measures may obscure. He chairs the epidemiology and health promotion department at NYU College of Dentistry.

Water access reveals a sharp divide

The findings connect dental health with both household conditions and access to care. Non-white populations had higher cavity risk across age groups. Private dental insurance and recent dental visits were associated with lower risk. Water source produced another notable difference among 12-year-olds in one analysis.

62.5%Without fluoridated water

In one analysis, 62.5% of 12-year-olds without access to fluoridated water had cavities.

35.5%With fluoridated water

Among 12-year-olds with fluoridated water systems, 35.5% had cavities.

Targeted prevention, with local limits

Hugo’s goal is to help public-health teams make better use of limited resources by targeting preventive care toward groups most likely to benefit. The study’s proposed role is identifying higher-risk population clusters from social conditions, rather than replacing an individual clinical diagnosis. That places its intended use in prevention planning.

The authors say the findings may extend to other middle-income countries, but they expect local conditions to change which variables explain risk. Access to dental care and benefits are among those differences. Their aim is to repeat the approach with country-specific factors, developing prevention strategies suited to each setting.

A U.S. analysis is already in preparation. Hugo and his colleagues plan to use data from the National Health and Nutrition Examination Survey. That is a planned extension of the work, not a finding that the Brazilian models perform equally well in a different country.