A novel machine learning model for class III surgery decision

• DOI: 10.1007/s00056-022-00421-7
• Publication Date: 2022-08-26T11:02:42Z
• Authors: Hunter Lee; Sunna Ahmad; Michael Frazier; Mehmet Murat Dundar; Hakan Turkkahraman
• Keywords: Male; Adolescent; Orthognathic Surgical Procedures; orthognathic surgery; Reproducibility of Results; artificial intelligence; Sensitivity and Specificity; Machine Learning; 03 medical and health sciences; 0302 clinical medicine; Humans; Original Article; Female; Neural Networks, Computer; computer-assisted decision making

Abstract Purpose The primary purpose of this study was to develop a new machine learning model for the surgery/non-surgery decision in class III patients and evaluate validity reliability model. Methods sample consisted 196 skeletal patients. All cases were allocated randomly, 136 training set remaining 60 test set. Using set, success rate artificial neural network estimated, along with 95% confidence interval. To predict surgical cases, we trained binary classifier using two different methods: random forest (RF) logistic regression (LR). Results Both RF LR showed high separability when classifying each patient or non-surgical treatment. achieved an area under curve (AUC) 0.9395 on intervals computed by bootstrap sampling as lower bound = 0.7908 higher 0.9799. On other hand, AUC 0.937 0.8467 0.9812. Conclusions models can be used generate accurate reliable algorithms successfully classify up 90%. features selected coincide clinical that clinicians weigh heavily determining treatment plan. This further supports overjet, Wits appraisal, incisor angulation, Holdaway H angle strong predictors assessing patient’s needs.