Developing and validating a machine learning system for Orthodontic treatment planning - A descriptive study

Abstract

Background : The success of any orthodontic treatment involves a robust clinical examination along with a comprehensive treatment planning that involves a combination of various essential factors. Orthodontic treatment is a long‐term procedure and accurate diagnosis and treatment planning forms the essence of successful orthodontic treatment. The advent of AI (artificial intelligence) has revolutionized dentistry and can prove to be a valuable tool in orthodontic treatment planning. Hence the objective of this study was to construct a machine learning predictive model for orthodontic treatment planning and correlate the treatment plan prediction of the model to treatment plan decided by expert Orthodontists. Methodology: The sample consisted of 650 case records of orthodontically treated patients satisfying inclusion criteria. The data was split into a training and a test set. The input layer of variables consisted of parameters that most commonly determine the diagnosis and treatment planning of the patient. The output layer comprised of various modes of treatment. The test set was used to check for the efficacy of the ML predicted treatment plan and compared to that of the decision made by the expert Orthodontists. Results: The ML models were trained and the accuracy for Treatment predictive models was deduced. The algorithm Random Forest provided the best results and was used as the predictive model for Treatment planning. This model showed an accuracy of 94.53% and an F1 score of 96.75% for treatment objectives. Conclusion: The use of an automated machine learning system allows the generation of orthodontic treatment predictive models.