Artificial intelligence in dentistry: High-accuracy models for caries risk assessment

Summaries 2 min 446

Font Size

Created On: 24/03/2026

Dental caries is a widespread chronic oral disease that demands timely intervention to prevent complications and minimize healthcare costs

See All

Key take away

Machine learning–based caries prediction models provide fast, reliable, and standardized risk scores with >95% accuracy.

Background

Dental caries is a widespread chronic oral disease that demands timely intervention to prevent complications and minimize healthcare costs. Accurate caries risk assessment plays a fundamental role in prevention; however, conventional methods rely heavily on expert clinical judgment, limiting scalability and consistency.

Machine learning in dentistry yields a scalable, efficient, and standardized approach to improve risk prediction. This study sought to develop and validate advanced machine learning models for caries risk prediction in adults, boosting accuracy and clinical decision-making.

Method

A systematic review identified 7 key caries risk predictors, which were applied to a balanced dataset of 3,000 patient records from Universiti Teknologi MARA, categorized into low, moderate, and high-risk groups. Overall, 7 machine learning algorithms—logistic regression, decision tree, random forest, XGBoost, k-nearest neighbors, multi-layer perceptron, and support vector machine—were optimized via K-fold cross-validation and stacking ensemble techniques to improve predictive performance.

Result

The stacked ensemble model (two-model approach) delivered the highest performance, attaining 95.17% accuracy and an exceptional receiver operating characteristic - area under the curve (ROC-AUC) of 99.78%, indicating superior predictive capability. The three-model stacking approach also illustrated strong results, with 93.63% accuracy and 96.82% specificity. Among individual models, random forest (90.47%) and XGBoost (90.20%) exhibited robust classification accuracy, highlighting their potency in caries risk prediction.

Conclusion

Ensemble machine learning models substantially improved the accuracy and reliability of dental caries risk assessment, minimizing dependence on subjective clinical evaluation. These data-driven approaches enabled standardized risk prediction, support clinical workflow integration, and enhance preventive dentistry strategies and overall patient care.