The broad term artificial intelligence (AI) refers to the capability of machines and computer systems to perform tasks that typically require human intelligence [1]. The field of AI focuses on developing intelligent systems that exhibit cognitive functions such as language comprehension, learning, reasoning, and problem-solving [2]. Several subfields of AI have been widely applied in biological and medical diagnostics, including machine learning (ML), artificial neural networks (ANNs), convolutional neural networks (CNNs), and deep learning (DL) [2].

In dentistry, AI has emerged as a powerful tool with applications ranging from administrative tasks—such as scheduling and organizing appointments—to supporting clinical diagnosis and treatment planning. AI systems can perform many tasks in dental practice with greater precision, efficiency, and consistency compared to human performance [2]. With the introduction of digital technologies such as intraoral scanners, cone-beam computed tomography (CBCT), and advanced imaging software, orthodontics has experienced remarkable technological progress [3]. The success of orthodontic treatment largely depends on accurate diagnosis and comprehensive treatment planning, which are traditionally based on the patient’s medical and dental records, clinical examination, study models, and cephalometric radiographs [4].

AI technologies can assist orthodontic specialists in various diagnostic and decision-making processes, including cephalometric landmark identification, assessment of the need for orthodontic extractions, evaluation of cervical vertebral maturation, prediction of facial attractiveness following orthognathic surgery, and overall orthodontic treatment planning [5].

The purpose of this systematic review was to evaluate the validity and scope of AI-based models that have been implemented in orthodontic diagnostics.

In orthodontics, accurate diagnosis is crucial because it directly influences treatment planning and the resulting clinical outcomes. Recently, there has been growing interest among orthodontists in incorporating AI-based diagnostic tools into treatment planning [20]. AI models offer the potential to enhance patient care by supporting clinicians in decision-making and improving workflow efficiency, thereby saving time and optimizing clinical care [21]. The primary aim of this systematic review was to evaluate the validity and performance of AI-based models across various orthodontic diagnostic applications.

A key aspect of assessing treatment effectiveness is the evaluation of the patient’s facial appearance [21]. AI has shown particular promise in this area. An artificial neural network (ANN) model was developed to predict post-orthognathic surgery facial profiles, demonstrating encouraging results with prediction accuracy exceeding 80% [21]. Similarly, Patcas et al. [14] assessed the facial aesthetics of cleft patients using AI, finding that the model’s outcomes were comparable to the mean scores assigned by laypeople, orthodontists, and oral surgeons. These findings suggest that AI-based systems can provide objective and reproducible assessments, which may support clinical decision-making and treatment planning [14][21]. Nonetheless, the authors emphasized that further development is necessary to enhance the predictive accuracy of these systems [14][21].

Beyond aesthetics, AI is increasingly employed to help practitioners determine the need for orthodontic treatment. In recent years, AI-based clinical decision support systems have been implemented to assist clinicians in making informed choices [22]. For instance, Thanathornwong et al. [10] demonstrated that a Bayesian probabilistic model could classify patients accurately into those requiring or not requiring orthodontic treatment, based on intra- and extra-oral data. The Bayesian Network (BN) serves as the foundation for such decision support systems by modeling the causal relationships among multiple factors influencing treatment necessity [10].

Additionally, ANN-based models have been applied to support decisions regarding orthodontic extractions [23]. These systems often rely on template-matching approaches, where the patient’s data are compared with a comprehensive reference database to identify similar cases [9][23]. This allows AI expert systems to simulate the decision-making process of experienced specialists, providing guidance for less experienced clinicians while leaving the final decision to the practitioner [23]. An added advantage of AI is its flexibility; different diagnostic philosophies can be encoded into expert systems, allowing for diverse approaches to treatment planning [9][23].

When orthodontic treatment is indicated, multiple additional factors must be considered, including the patient’s biological maturity [24]. Accurate assessment of biological maturity is critical for timing treatment interventions, particularly in growing patients [12][17][24]. In recent years, evaluation of cervical vertebral morphology and development has been considered one of the most reliable methods for assessing skeletal maturity [24]. Uysal et al. [25] demonstrated that cervical vertebral stages could be used clinically to estimate growth and development, establishing correlations between chronological age and skeletal age assessed through hand-wrist and cervical vertebral radiographs using Spearman rank-order correlation coefficients. Artificial intelligence has been increasingly applied to facilitate the identification of cervical vertebral growth stages [12]. Kok et al. developed an ANN algorithm capable of reliably determining all stages of cervical vertebral maturation, highlighting the potential of AI to provide more precise and unbiased assessments in clinical practice [12]. By reducing subjective variability inherent in human assessment, AI allows for standardized evaluations that may improve treatment planning and outcomes. In addition to skeletal maturity, prediction of mandibular growth is a critical component of orthodontic diagnosis and treatment planning [26]. Mathematical models have been developed to forecast mandibular growth in children based on population growth curves [26]. Buschang et al. [26] compared a population-based model with mean annual growth velocities, using multilevel modeling to account for individual and measurement-level differences. Their predictions achieved an accuracy of 76% - 77% for both males and females. More recently, Wood et al. [17] applied machine learning techniques to predict post-pubertal mandibular length and Y-axis growth in males, achieving remarkably high accuracy, ranging from 95.80% to 97.64% for mandibular length and 96.60% to 98.34% for Y-axis growth. These findings emphasize that precise prediction of growth patterns can substantially influence treatment strategies for skeletal malocclusions [17]. Clinicians alone may find it challenging to process the extensive data required for such predictions, highlighting the advantage of AI-assisted analysis in handling complex datasets and providing reliable, evidence-based growth forecasts [17].

AI has also demonstrated considerable utility in cephalometric analysis, a cornerstone of orthodontic diagnosis and treatment planning [27]. Accurate identification of cephalometric landmarks underpins measurements of craniofacial angles, distances, and ratios, which in turn guide diagnostic and therapeutic decisions [28]. Traditionally, the quality of cephalometric analyses has depended heavily on the experience of the clinician, leading to variability and potential errors [28]. AI algorithms offer a promising solution by providing automated, reproducible landmark identification [15][16][27]. Bourzgui et al. [29] developed an automated 3D cephalometric point detection system, achieving a mean error of 2.32 mm across 21 anatomical points using both anatomical and geometric knowledge. Similarly, Kunz et al. [16] reported that AI-driven cephalometric analyses demonstrated clinical precision comparable to human examiners across 12 orthodontic parameters, with no statistically significant differences from the gold standard. Hwang et al. [15] corroborated these findings, showing that AI systems can match human accuracy in landmark identification. These studies indicate that AI-based methods can serve as reliable adjuncts for repeated analyses, enhancing efficiency and consistency in clinical practice [15][16][27]-[29].

Once a diagnosis is established, clinicians must formulate a treatment plan tailored to the patient’s needs [18]. In cases where orthodontic treatment alone is insufficient, orthognathic surgery may be required [11]. Optimal outcomes depend on careful preoperative orthodontic preparation to address dental compensations, which maximizes the effectiveness of surgical interventions [30]. Experienced clinicians often draw upon accumulated clinical experience to develop individualized treatment philosophies, which are inherently complex and challenging to transfer to less experienced practitioners within a short period of time [11][30]. AI offers support in such scenarios by simulating expert decision-making processes [31]. Choi et al. [11] developed an AI model capable of making surgery versus non-surgery and extraction decisions. The model demonstrated high predictive performance, with success rates of 95% for the training set, 97% for the validation set, 96% for the test set, and 96% overall for surgery/non-surgery decisions, while correctly predicting the type of surgery in 100% of cases. Such AI tools provide an additional layer of knowledge, particularly in borderline cases, while maintaining clinician oversight and decision-making authority [13].

Integrating AI into treatment planning may therefore enhance decision-making efficiency, support less experienced clinicians, and contribute to improved patient outcomes, while ensuring that human expertise remains central to the management of complex orthodontic and orthognathic cases [11][13].

The implementation of artificial intelligence has markedly advanced the precision of diagnostic processes and treatment planning within the medical domain. These systems have demonstrated exceptional efficacy in performing their designated functions. In orthodontics, where clinical decision-making is fundamentally driven by diagnostic assessment, the integration of AI technologies has proven to be both effective and advantageous, contributing to enhanced diagnostic accuracy and more informed therapeutic planning.

This study is a systematic review of previously published studies and did not involve direct research on human participants; therefore, ethics committee approval was not required.

Both authors contributed equally to all stages of this work, from the study conception and design to data collection, analysis, manuscript drafting, and final approval.