Sara Haghighat

Artificial intelligence (AI) is increasingly being integrated into orthodontic research and clinical practice. This review paper consists of three parts and aims to establish a foundational understanding of AI concepts for readers from non‐technical backgrounds while familiarising them with its applications and challenges within the field of orthodontics. The first part provides a comprehensive introduction to the basic concepts of AI, including definitions of machine learning and deep learning models, alongside various learning types: Supervised, unsupervised, and reinforcement learning. To explain how learning occurs and how to optimise it during model training, key features of the “learning” process are introduced. A detailed overview of supervised learning is presented, including data preparation, model selection, and performance evaluation. Following this introduction, different architectures and algorithms of machine learning and neural networks are examined, with examples of their applications in orthodontics, including benefits for clinical decision‐making, image processing, and patient monitoring. Finally, this review will consider innovations in AI that hold promise for enhancing orthodontic practice, such as the use of synthetic data, transformers, generative models, and large language models.

Impaction of the mandibular third molar in proximity to the mandibular canal increases the risk of inferior alveolar nerve injury. Panoramic radiography is routinely used to assess this relationship. Automated classification of molar–canal overlap could support clinical triage and reduce unnecessary CBCT referrals, while Federated Learning (FL) enables multi-center collaboration without sharing patient data. We compared Local Learning (LL), FL, and Centralized Learning (CL) for binary overlap/no-overlap classification on cropped panoramic radiographs partitioned across eight independent labelers in a simulated heterogeneous multi-center setting. A pretrained ResNet-34 was trained under each paradigm and evaluated using per-client metrics with locally optimized thresholds and pooled test performance with a global threshold. Performance was assessed using area under the receiver operating characteristic curve (AUC) and threshold-based metrics, alongside training dynamics, Grad-CAM visualizations, and server-side aggregate monitoring signals. On the test set, CL achieved the highest performance (AUC 0.831; accuracy ≈ 0.782), FL showed intermediate performance (AUC 0.757; accuracy ≈ 0.703), and LL generalized poorly across clients (AUC range ≈ 0.619–0.734; mean ≈ 0.672). Training curves suggested overfitting, particularly in LL models, and Grad-CAM indicated more anatomically focused attention in CL and FL. Overall, centralized training provided the strongest performance, while FL offers a privacy-preserving alternative that outperforms LL.

Obstructive sleep apnea (OSA) is a highly prevalent sleep disorder. Misdiagnosis might lead to several systemic conditions, including hypertension, vascular damage, and cognitive impairment. The gold-standard diagnostic tool for OSA is polysomnography, which is expensive, time-consuming, and not accessible everywhere. Artificial intelligence (AI) algorithms can facilitate diagnosis by detecting patients’ signs and symptoms. In this systematic review, we evaluated the diagnostic accuracy of AI models in detecting sleep apnea. We searched six major databases, PubMed^®, Cochrane, Web of Science, Scopus, Embase, and IEEE Xplore, using keywords related to AI and OSA. Eligible studies focused on adult populations, used in-laboratory PSG as the reference standard, and applied AI models trained on multiple clinical features. Reviews, pediatric studies, and articles lacking accuracy metrics were excluded. From the included articles, data were extracted regarding patients and datasets, type of AI model applied, accuracy report, and explainability of the AI model. A risk of bias assessment was done using the QUADAS-2 checklist. Thirteen studies were included in our final analysis. The AI models consisted of deep learning, machine learning, and hybrid models with various architectures. The reported accuracy of studies ranged from 67.03 to 98.6%, with the highest being related to hybrid and deep learning models. Risk of bias assessment showed that 7 of the studies had a low risk of bias, indicating high reliability. AI-driven models, particularly deep learning and hybrid architectures, show significant promise in diagnosing obstructive sleep apnea. However, challenges such as transparency, explainability, and variability in performance necessitate diverse training datasets to improve generalizability for clinical adoption. The protocol of this systematic review was registered in PROSPERO (CRD42023453789), available from: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023453789. The online version contains supplementary material available at 10.1186/s12911-025-03129-x.

Dental pulp derived-mesenchymal stem cells (DP-MSCs) is considered a suitable are candidate for tissue engineering techniques and osseous reconstruction. Based on the hypothesis that Hypericum perforatum, Elaeagnus Angustifolia and Psidium guajava extracts can be used in cell-based bone tissue engineering due to meagre cytotoxicity response in the cell culture medium, their effects on the viability and metabolic activity of DP-MSCs were investigated and compared with each extract. DP-MSCs were extracted from human dental pulp, characterized by flow cytometry, and differentiated into Osteogenic and adipogenic lineages which were then cultured in different concentrations of E. Angustifolia, H. perforatum and P. guajava extracts at different time intervals followed by MTT assay evaluation. The dental pulp mesenchymal stem cells were evaluated for their plastic adherence ability, fibroblast-like and spindle morphology. According to flow cytometry data, isolated cells from DP-MSCs expressed MSCs markers. A comparison of herbal extracts’ concentrations revealed that 500 μg/ml was toxic to dental pulp stem cells, a guide to the toxic dose for DP-MSCs. The P.guajava bore low toxicity and increased dental pulp stem cell viability in comparison to the other two herbal extracts. The hydro-alcoholic extracts of E. Angustifolia, H. perforatum, and P. guajava were efficient in DP-MSCs viability, and therefore were concluded to be useful in maintaining structural and functional cell viability. It was also concluded that the co-culture of stem cells with herbal elements could stimulate endogenous factors to enhance the proliferation and viability of MSCs. Graphic Abstract

Introduction. Oral lichen planus (OLP) is a chronic mucocutaneous disease. It is mainly an immune system‐related disorder. Vitamins can modulate immune system functions, and thus, vitamin deficiency might have roles in exacerbating OLP. We aim to determine the serum levels of vitamins A, B12, C, D3, and E in OLP patients. Methods and Materials. 34 OLP patients referred to Shiraz Dental School entered the study. Blood samples were collected and levels of A, B12, C, D3, and E vitamins were measured in serum. 43 healthy people were also included as the control group. Serum levels of vitamins were measured by HPLC (A, B12, D3, and E) and Kiazist analyzing kit (vitamin C). Results. Most of the patients were female (62.3%), and the mean age of patients was 48.03 ± 11.57. Serum levels of vitamins A, C, and E were lower in OLP patients in comparison with the healthy group; however, the difference was not significant. Vitamins B12 and D3 were higher in the OLP group but the difference was not significant. Conclusion. Serum levels of vitamins A, B12, C, D3, and E do not have a significant difference in OLP patients and healthy groups. These vitamins may not have a considerable role in OLP pathogenesis in the southwest of Iran.