Table 8.

AI outcomes in orthodontics.

Target	AI Model	Sample	Results	Study
To decide if extractions are necessary prior to orthodontic treatment	Backpropagation ANN	Data from 180 patients	Accuracy of 80% in predicting whether extraction or non-extraction treatments seemed appropriate for malocclusion patients aged 11 to 15 years.	Xie et al., (2010) [183]
To decide if extractions are necessary prior to orthodontic treatment	ANN	In total, 12 cephalometric variables and 6 indexes from 156 patients	Accuracy of 92%	Jung and Kim (2016) [184]
Determination of growth and development by cervical vertebrae stages	ANN	Cephalometric radiographs from 300 subjects	Mean accuracy of 77.2%	Kök et al., (2019) [187]
Osteoarthritis of the temporomandibular joint diagnosis	Logistic regression, random forest, LightGBM, XGBoost	CBCT blood serum saliva clinical investigation	Accuracy of 82.3%	Bianchi et al., (2020) [188]
Determination of growth and development periods	ANN	Cephalometric and hand–wrist radiographs in 419 subjects	Accuracy of 4.27%	Kök et al., (2021) [189]

AI, artificial intelligence; ANN, artificial neural network; CBCT, cone-beam computed tomography.