Table 5.

Characteristics of the studies.

Authors Years	Type of Study	Type of AI	Materials and Methods	Results
Patcas et al., 2019 [32]	Observational study	AI to explain how orthodontic therapy affects facial beauty and apparent age.	Every photograph has patient-related information (patient age, sex, malocclusion, and surgeries performed) tagged on it. With specialized CNNs trained on >0.5 million photos for age estimation and with >17 million attractiveness ratings, face attractiveness (score: 0–100) and apparent age were determined for each image.	The algorithms discovered that the vast majority of patients’ looks improved following therapy (66.4%), leading to a roughly one-year younger appearance, especially after profile-altering surgery. Similar positive effects of orthognathic therapy on beauty were seen in 74.7% of cases, particularly following lower jaw surgery.
Caruso et al., 2021 [68]	Case report	Correct biomechanics were guaranteed by the software’s analysis of the aligner’s fit and retention.	Depending on the chosen protocol, guided scanning will transmit 20–30 photos to the servers for processing, which may be broken down into four steps: Step 1: The system processes the raw photos. They are evaluated for quality to see if the patient requires another scan or not; Step 2: Using a prediction score (% of certainty), the algorithm can locate teeth and identify them. In some orthodontic extraction situations, the technology is so sophisticated that it can sometimes tell if a tooth is a first or second premolar. Additionally, the gingiva is shown; Step 3: Finding the various clinical parameters; Step 4: The AI will review the data and, using the selected strategy, will provide instructions to the patient and the team.	The patient demonstrated exceptional compliance with and confidence in the DM system on receiving nearly all “GO” signals during his therapy. Up to the completion of all clinical objectives, monitoring was activated. Therefore, it was just put on hold while awaiting the new aligners.
Lee et al., 2022 [70]	Clinical Study	To compare the creation of integrated tooth models (ITMs) with the manual method and to assess the accuracy of DL-based ITMs by combining intraoral scans and CBCT scans for three-dimensional (3D) root position evaluation during orthodontic treatment.	15 patients who underwent orthodontic treatment with premolar extraction had intraoral scans and related CBCT scans taken before and after treatment.	The procedure times taken to obtain the measurements were longer in the manual method than in the DL method.
Ferlito et al., 2023 [72]	Prospective study	Clear aligner treatment has lately gained popularity due to the use of AI for remote monitoring. DL algorithms on a patient’s mobile smartphone were used to decide readiness to move to the next aligner (i.e., “GO” versus “NO-GO”) and detect places where the teeth do not match the clear aligners.	Thirty patients under treatment with clear aligners at an academic clinic were scanned twice using a remote smartphone monitoring software, and the results were compared.	A 44.7% gauge compatibility was observed. Between Scan 1 and 2, 83.3% of patient instructions agreed; however, 0% agreed on whether and/or how many teeth had tracking difficulties. In the mesiodistal, buccolingual, occlusogingival, tip, torque, and rotational dimensions, patients who received a “GO” instruction exhibited mean differences of 1.997 mm, 1.901 mm, 0.530 mm, 8.911, 7.827, and 7.049, respectively. These differences were not statistically significant when patients were given “NO-GO” instructions.