Table 4.

Advanced AI applications in orthodontics.

Authors	Summarized Abstract	Methods Used	Results	Conclusions
Niño-Sandoval [35]	-Predicted mandibular morphology using automated learning techniques in Colombian patients.	-Automated learning techniques: artificial neural networks and support vector regression. -Support vector regression (SVR) and artificial neural networks (ANNs).	-Coefficients ranged from 0.84 to 0.99 with artificial neural networks. -Support vector regression achieved two coefficients above 0.7.	-Automated learning techniques predict mandibular morphology accurately. -Craniomaxillary variables can be used for facial reconstruction.
Panesar [36]	-AI improved precision and accuracy of cephalometric analyses significantly.	-Deep learning AI with and without human augmentation.	-AI improved precision and accuracy in cephalometric analyses significantly.	-AI/human augmentation method enhances less experienced dental professionals’ performance.
Alessandri-Bonetti [37]	-AI-assisted cephalometric analysis compared to manual software, showing reliability.	-Dahlberg equation for intra-and inter-operator reliability in cephalometric parameters.	-No significant difference in intra-and inter-operator measurements in cephalometric parameters. -Higher errors observed in posterior facial height and facial axis angle.	-Fully automated AI-assisted cephalometric software shows reliable and accurate measurements. -Digital advances cannot replace the orthodontist’s role in diagnosis.
Kochhar [38]	-Evaluated bone volume reliability in cleft lip and palate patients.	-Evaluation of bone volume using OsiriX software in CBCT scans. -Assessment of bone volume reliability by three specialists.	-Left-side clefts required more bone volume than the right side. -Age and gender did not correlate with bone volume needed.	-OsiriX software shows good reliability in bone volume measurements.
Çoban [39]	-Compared DM and AI cephalometric analysis in different skeletal malocclusions.	-Digital manual cephalometric analysis with Dolphin Imaging software (v. 11.5, California, USA). -AI-based cephalometric analysis using the WebCeph platform.	-Significant differences in most parameters between digital manual and AI methods.	-AI method needs further development for specific malocclusions. -Both AI and manual methods are suitable for orthodontic analysis.
Jiang [40]	-AI system for cephalometric analysis with automated landmark localization proposed.	-Collection of 9870 cephalograms from 20 medical institutions for training. -Development of a two-stage convolutional neural network for landmark localization.	-Average landmark prediction error was 0.94 ± 0.74 mm. -System achieved an average classification accuracy of 89.33%.	-Proposed AI system for cephalometric analysis improves diagnostic efficiency.
Li [41]	-ANN predicted orthodontic treatment plans with high accuracy and feasibility. -Study explores RBES, CBES, and ANN for orthodontic treatment planning.	-Multilayer perceptron artificial neural networks for orthodontic treatment planning. -Rule-based expert systems and case-based expert systems were utilized.	-Neural network accuracy: 94.0% for extraction–nonextraction prediction. -Extraction patterns accuracy: 84.2%. Anchorage patterns accuracy: 92.8%.	-Artificial neural networks aid in accurate orthodontic treatment planning. -MLPs show high accuracy in predicting extraction–nonextraction, extraction, and anchorage patterns.
Silva [42]	-CEFBOT (RadioMemory Ltd., Belo Horizonte, Brazil) AI software reliable for cephalometric landmark annotation and measurements.	-Cephalometric landmark annotation and measurements using AI-based software. -Duplicate measurements by human examiner and CEFBOT for reliability assessment.	-Frankfurt horizontal plane–true horizontal line angular measurement had the lowest reproducibility. -CEFBOT was unable to measure the distance from the glabella to the subnasale.	-No statistically significant difference between human examiner and CEFBOT measurements. -AI-based CEFBOT is comparable to human examiners in reproducibility.