New AI Model Predicts Who Is Most Likely To Develop Cavities Before They Start

AI Models Uncover Hidden Patterns in Global Tooth Decay Risk

A team of researchers is turning to artificial intelligence to tackle the most pervasive health issue on the planet: dental cavities. By shifting the focus from individual clinical diagnostics to broader social and environmental factors, scientists are developing predictive models that can identify population groups at the highest risk for tooth decay.

The findings, recently detailed in the journal Caries Research, demonstrate how machine learning can process complex, large-scale data to reveal how structural and social inequalities drive dental disease. According to the study authors, this approach offers a more precise, high-resolution view of public health, allowing authorities to direct preventive resources where they are needed most.

To train and test these models, researchers utilized data from the 2023 Brazilian Oral Health Survey, which included clinical and demographic information from 23,726 participants spanning multiple age groups. The dataset provided a comprehensive look at the intersection of oral health and social determinants, ranging from household density and wealth to access to fluoridated water and insurance coverage.

The analysis highlighted significant disparities. Among 12-year-olds, for example, those living without access to fluoridated water saw cavity rates of 62.5 percent, compared to 35.5 percent for those with access to fluoridated systems. The study also identified that non-white populations, as well as those living in high-density households with lower wealth, faced a statistically higher risk of developing tooth decay.

While traditional statistical methods can identify these general trends, the research team discovered that machine learning algorithms were uniquely capable of uncovering the subtle, multidimensional interactions between these variables. By analyzing how different factors—such as the time since a person’s last dental visit or their specific socioeconomic status—interact, the AI can detect risk profiles that might remain hidden within community-wide averages.

“Integrating machine learning into this framework goes beyond merely predicting cavity risk; it addresses the complexity and multidimensional interactions between social determinants that traditional measures often mask,” Hugo explains. “Our machine learning approach acts as a high-resolution sensor that identifies individuals whose profiles signal a heightened risk that would otherwise be invisible within community averages.”

Looking ahead, the team aims to expand this work beyond Brazil. They are currently preparing a similar analysis using data from the U.S. National Health and Nutrition Examination Survey. The researchers hope that by customizing the models to account for the specific social and economic realities of different nations, they can help policymakers design more effective, evidence-based prevention strategies.

Ultimately, the goal is to move toward a more efficient public health model. By identifying vulnerable clusters before decay becomes severe, health departments can optimize their spending and ensure that limited resources generate the greatest possible impact for the public. “When we think about public health and preventive interventions, we need to be accountable to society and make the best use of the resources we have,” says Hugo.