Prediction of patient cooperation with artificial intelligence before orthodontic treatment

Aim: The aim of this study is to test whether it can predict the patient's cooperation during orthodontic treatment with artificial intelligence before starting the treatment. Methods: According to the orthodontic patient cooperation scale that their physicians filled out in the 12th month of their treatment, 230 patients who had fixed orthodontic treatment in our clinic were split into two distinct groups cooperative and non-cooperative. Voice recordings, handwriting samples and facial photographs of the patients were obtained. These collected data were analyzed with various machine learning algorithms and artificial neural network models, and the relationship of facial photographs, handwriting, and voice with patient cooperation and the predictability of patient cooperation before starting treatment were investigated. The learning transfer technique was used to increase the performance and success rates of the models. Results: With a success percentage of 66.0%, Xception and DenseNet were the models that performed the best among those that assessed the face pictures. Among the models used to evaluate handwriting samples InceptionResNet (72.0%) and NasNetMobile (70.0%) were the best performing models. The algorithms that performed best in the assessment of sound data were the ESA model together with Linear Discriminant Analysis, K-Nearest Neighbor, Support Vector Machine, Extra Tree Classifier, and Stacking Classifier, with a success percentage of 57.0%. Conclusions: The physiognomy features of the face, features of handwriting, tone and frequency of the voice can be used for cooperation prediction. Handwriting; It carries more information about cooperation than face and voice. The highest success rate obtained with artificial neural networks in the evaluation of handwriting samples, face photographs and voice recordings was 72.0%, 63.0% and 57.0%, respectively. Keywords: Artificial intelligence, Convolutional neural networks, Cooperation