Development of a recommender system for dental care using machine learning

Resource mismanagement along with the underutilization of dental care has led to serious health and economic consequences. Artificial intelligence was applied to a national health database to develop recommendations for dental care. The data were obtained from the 2013–2014 National Health and Nutrition Examination Survey to perform machine learning. Feature selection was done using LASSO in R to determine the best regression model. Prediction models were developed using several supervised machine learning algorithms, including logistic regression, support vector machine, random forest, and classification and regression tree. Feature selection by LASSO along with the inclusion of additional clinically relevant variables identified 8 top features associated with recommendation for dental care. The top 3 features include gum health, number of prescription medications taken, and race. Gum health shows a significantly higher relative importance compared to other features. Demographics, healthcare access, and general health variables were identified as top features related to receiving additional dental care, consistent with prior research. Practicing dentists and other healthcare professionals can follow this model to enable precision dentistry through the incorporation of our algorithms into computerized screening tool or decision tree diagram to achieve more efficient and personalized preventive strategies and treatment protocols in dental care.