Health Data Sci. 2025 Apr 1;5:0256. doi: 10.34133/hds.0256. eCollection 2025.

ABSTRACT

Background: Multimodal large language models (LLMs) have shown potential in various health-related fields. However, many healthcare studies have raised concerns about the reliability and biases of LLMs in healthcare applications. Methods: To explore the practical application of multimodal LLMs in skin disease identification, and to evaluate sex and age biases, we tested the performance of 2 popular multimodal LLMs, ChatGPT-4 and LLaVA-1.6, across diverse sex and age groups using a subset of a large dermatoscopic dataset containing around 10,000 images and 3 skin diseases (melanoma, melanocytic nevi, and benign keratosis-like lesions). Results: In comparison to 3 deep learning models (VGG16, ResNet50, and Model Derm) based on convolutional neural network (CNN), one vision transformer model (Swin-B), we found that ChatGPT-4 and LLaVA-1.6 demonstrated overall accuracies that were 3% and 23% higher (and F1-scores that were 4% and 34% higher), respectively, than the best performing CNN-based baseline while maintaining accuracies that were 38% and 26% lower (and F1-scores that were 38% and 19% lower), respectively, than Swin-B. Meanwhile, ChatGPT-4 is generally unbiased in identifying these skin diseases across sex and age groups, while LLaVA-1.6 is generally unbiased across age groups, in contrast to Swin-B, which is biased in identifying melanocytic nevi. Conclusions: This study suggests the usefulness and fairness of LLMs in dermatological applications, aiding physicians and practitioners with diagnostic recommendations and patient screening. To further verify and evaluate the reliability and fairness of LLMs in healthcare, experiments using larger and more diverse datasets need to be performed in the future.

PMID:40170800 | PMC:PMC11961048 | DOI:10.34133/hds.0256

Eur Rev Med Pharmacol Sci. 2025 Mar;29(3):123-134. doi: 10.26355/eurrev_202503_37125.

ABSTRACT

The interstitial lung disease (ILD) field is rapidly expanding as new insights highlight novel mechanisms and procedures that influence epidemiology, diagnosis, and treatment. The aim of this review is to report on recent advancements and future perspectives in clinical management and research in ILD, particularly in idiopathic pulmonary fibrosis (IPF), for the most common ILD. Whilst high-resolution computed tomography (HRCT) remains the gold standard for diagnosis, we focus on newer diagnostic techniques, including IPF genome analysis and epigenetics, biomarkers, bronchoscope robotic navigation, and transbronchial lung cryo biopsy for improving diagnostic accuracy. Further, we report IPF associated with pulmonary hypertension Group 3 and scores for defining disease progression. Positron emission tomography/computed tomography, treatment with prostacyclin and antifibrotic drugs, and lung transplantation as potential treatments for end-stage IPF are discussed. Lastly, we discuss contemporary perspectives on interstitial lung abnormalities (ILA), IPF associated with lung cancer, and the use of artificial intelligence (AI) for ILD diagnosis and monitoring.

GRAPHICAL ABSTRACT: https://www.europeanreview.org/wp/wp-content/uploads/123-134.jpg.

PMID:40171787 | DOI:10.26355/eurrev_202503_37125

iScience. 2025 Mar 4;28(4):112146. doi: 10.1016/j.isci.2025.112146. eCollection 2025 Apr 18.

ABSTRACT

Epidermal growth factor receptor (Egfr)-driven signaling regulates fundamental homeostatic processes. Dysregulated signaling via Egfr is implicated in numerous disease pathologies and distinct Egfr-associated disease etiologies are known to be tissue-specific. The molecular basis of this tissue-specificity remains poorly understood. Most studies of Egfr signaling to date have been performed in vitro or in tissue-specific mouse models of disease, which has limited insight into Egfr signaling patterns in healthy tissues. Here, we carried out integrated phosphoproteomic, proteomic, and transcriptomic analyses of signaling changes across various mouse tissues in response to short-term stimulation with the Egfr ligand Egf. We show how both baseline and Egf-stimulated signaling dynamics differ between tissues. Moreover, we propose how baseline phosphorylation and total protein levels may be associated with clinically relevant tissue-specific Egfr-associated phenotypes. Altogether, our analyses illustrate tissue-specific effects of Egf stimulation and highlight potential links between underlying tissue biology and Egfr signaling output.

PMID:40171493 | PMC:PMC11960661 | DOI:10.1016/j.isci.2025.112146

RSC Adv. 2025 Apr 1;15(13):9884-9890. doi: 10.1039/d4ra07801b. eCollection 2025 Mar 28.

ABSTRACT

This study focuses on the creation and application of an advanced impedimetric immunosensor designed for the sensitive detection of lymphoma cancer cells. The sensor was developed by modifying a glassy carbon electrode (GCE) with gold nanoparticles (AuNPs) and 3,3'-dithiodipropionic acid di(N-hydroxysuccinimide ester) boronic acid (AuNPs@DTSP-BA), followed by the attachment of rituximab monoclonal antibody. Incorporating the boronic acid (BA) component enabled effective oriented immobilization of the antibody, thereby improving the performance of the biosensor. Various spectroscopic techniques were used to characterize the immunosensor. The developed immunosensor demonstrated the ability to detect lymphoma cancer cells across a wide linear range of 100 to 50 000 cells per mL, with a detection sensitivity of 64 cells per mL.

PMID:40171289 | PMC:PMC11959537 | DOI:10.1039/d4ra07801b

Cancer Discov. 2025 Apr 2;15(4):685-701. doi: 10.1158/2159-8290.CD-24-0861.

ABSTRACT

Viewing the hallmarks as a sequence of adaptations captures the "why" behind the "how" of the molecular changes driving cancer. This eco-evolutionary view distils the complexity of cancer progression into logical steps, providing a framework for understanding all existing and emerging hallmarks of cancer and developing therapeutic interventions.

PMID:40170539 | DOI:10.1158/2159-8290.CD-24-0861

Sheng Wu Gong Cheng Xue Bao. 2025 Mar 25;41(3):1152-1178. doi: 10.13345/j.cjb.250065.

ABSTRACT

Biomanufacturing has emerged as a crucial driving force for efficient material conversion through engineered cells or cell-free systems. However, the intrinsic spatiotemporal heterogeneity, complexity, and dynamic characteristics of these processes pose significant challenges to systematic understanding, optimization, and regulation. This review summarizes essential methodologies for multi-omics data acquisition and analyses for bioprocesses and outlines modelling approaches based on multi-omics data. Furthermore, we explore practical applications of multi-omics and modelling in fine-tuning process parameters, improving fermentation control, elucidating stress response mechanisms, optimizing nutrient supplementation, and enabling real-time monitoring and adaptive adjustment. The substantial potential offered by integrating multi-omics with computational modelling for precision bioprocessing is also discussed. Finally, we identify current challenges in bioprocess optimization and propose the possible solutions, the implementation of which will significantly deepen understanding and enhance control of complex bioprocesses, ultimately driving the rapid advancement of biomanufacturing.

PMID:40170317 | DOI:10.13345/j.cjb.250065

Sheng Wu Gong Cheng Xue Bao. 2025 Mar 25;41(3):1112-1132. doi: 10.13345/j.cjb.240966.

ABSTRACT

The metabolic reactions in cells, whether spontaneous or enzyme-catalyzed, form a highly complex metabolic network closely related to cellular physiological metabolic activities. The reconstruction of cellular physiological metabolic network models aids in systematically elucidating the relationship between genotype and growth phenotype, providing important computational biology tools for precisely characterizing cellular physiological metabolic activities and green biomanufacturing. This paper systematically introduces the latest research progress in different types of cellular physiological metabolic network models, including genome-scale metabolic models (GEMs), kinetic models, and enzyme-constrained genome-scale metabolic models (ecGEMs). Additionally, our paper discusses the advancements in the automated construction of GEMs and strategies for condition-specific GEM modeling. Considering artificial intelligence offers new opportunities for the high-precision construction of cellular physiological metabolic network models, our paper summarizes the applications of artificial intelligence in the development of kinetic models and enzyme-constrained models. In summary, the high-quality reconstruction of the aforementioned cellular physiological metabolic network models will provide robust computational support for future research in quantitative synthetic biology and systems biology.

PMID:40170315 | DOI:10.13345/j.cjb.240966

Sheng Wu Gong Cheng Xue Bao. 2025 Mar 25;41(3):1052-1078. doi: 10.13345/j.cjb.250061.

ABSTRACT

Biomanufacturing harnesses engineered cells for the large-scale production of biochemicals, biopharmaceuticals, biofuels, and biomaterials, playing a vital role in mitigating global environmental crises, achieving carbon peaking and neutrality, and driving the green transformation of the economy and society. The effective design and construction of these engineered cells require precise and comprehensive computational models. Recent technological breakthroughs including high-throughput sequencing, mass spectrometry, spectroscopy, and microfluidic devices, coupled with advances in data science, artificial intelligence, and automation, have enabled the rapid acquisition of large-scale biological datasets, thereby facilitating a deeper understanding of cellular dynamics and the construction of mechanism-based models with enhanced accuracy. This review systematically summarises the mathematical frameworks employed in cellular modelling. It begins by evaluating prevalent mathematical paradigms, such as network topology analyses, stochastic processes, and kinetic equations, critically assessing their applicability across various contexts. The discussion then categorises modelling strategies for specific cellular processes, including cellular growth and division, morphogenesis, DNA replication, transcriptional regulation, metabolism, signal transduction, and quorum sensing. We also examine the recent progress in developing whole-cell models through the integration of diverse cellular processes. The review concludes by addressing key challenges such as data scarcity, unknown mechanisms, multi-dimensional data integration, and exponentially escalating computational complexity. Overall, this work consolidates the mathematical models for the precise simulation of cellular processes, thereby enhancing our understanding of the molecular mechanisms governing cellular functions and contributing to the future design and optimisation of engineered organisms.

PMID:40170312 | DOI:10.13345/j.cjb.250061

Sheng Wu Gong Cheng Xue Bao. 2025 Mar 25;41(3):901-916. doi: 10.13345/j.cjb.240653.

ABSTRACT

Biomanufacturing is an advanced manufacturing method that integrates biology, chemistry, and engineering. It utilizes renewable biomass and biological organisms as production media to scale up the production of target products through fermentation. Compared with petrochemical routes, biomanufacturing offers significant advantages in reducing CO2 emissions, lowering energy consumption, and cutting costs. With the development of systems biology and synthetic biology and the accumulation of bioinformatics data, the integration of information technologies such as artificial intelligence, large models, and high-performance computing with biotechnology is propelling biomanufacturing into a data-driven era. This paper reviews the latest research progress on databases, knowledge bases, and large language models for biomanufacturing. It explores the development directions, challenges, and emerging technical methods in this field, aiming to provide guidance and inspiration for scientific research in related areas.

PMID:40170304 | DOI:10.13345/j.cjb.240653

Sheng Wu Gong Cheng Xue Bao. 2025 Mar 25;41(3):I-VIII. doi: 10.13345/j.cjb.250197.

ABSTRACT

Biomanufacturing is one of important strategies for sustainable development, China places significant emphasis on the development of biomanufacturing, and the national and local governments have successively introduced special policies for biomanufacturing, and vigorously developing biomanufacturing has become an unstoppable trend. At present, with the rapid development of systems biology and synthetic biology, biological big data and information technology are deeply integrating with biotechnology. Novel theories, methods and technologies for the design, creation and application of biological systems are constantly emerging, which promoted the development of biomanufacturing into the era of artificial intelligence (AI). In order to grasp the innovation and development of AI-driven biomanufacturing, we publish this special issue to review the opportunities, challenges, and development status of AI-driven biomanufacturing from aspects such as AI-driven enabling technologies, intelligent design and construction of biological parts, circuits and artificial cells, as well as intelligent bioprocess control and optimization, and look forward to the future developments. This will provide valuable references for effectively promoting technological innovation and industrial development in the field of biomanufacturing.

PMID:40170303 | DOI:10.13345/j.cjb.250197

Clin Transl Med. 2025 Apr;15(4):e70290. doi: 10.1002/ctm2.70290.

ABSTRACT

Single-cell omics has emerged as a powerful tool for elucidating cellular heterogeneity in health and disease. Parallel advances in artificial intelligence (AI), particularly in pattern recognition, feature extraction and predictive modelling, now offer unprecedented opportunities to translate these insights into clinical applications. Here, we propose single-cell -omics-based Disease Predictor through AI (scDisPreAI), a unified framework that leverages AI to integrate single-cell -omics data, enabling robust disease and disease-stage prediction, alongside biomarker discovery. The foundation of scDisPreAI lies in assembling a large, standardised database spanning diverse diseases and multiple disease stages. Rigorous data preprocessing, including normalisation and batch effect correction, ensures that biological rather than technical variation drives downstream models. Machine learning pipelines or deep learning architectures can then be trained in a multi-task fashion, classifying both disease identity and disease stage. Crucially, interpretability techniques such as SHapley Additive exPlanations (SHAP) values or attention weights pinpoint the genes most influential for these predictions, highlighting biomarkers that may be shared across diseases or disease stages. By consolidating predictive modelling with interpretable biomarker identification, scDisPreAI may be deployed as a clinical decision assistant, flagging potential therapeutic targets for drug repurposing and guiding tailored treatments. In this editorial, we propose the technical and methodological roadmap for scDisPreAI and emphasises future directions, including the incorporation of multi-omics, standardised protocols and prospective clinical validation, to fully harness the transformative potential of single-cell AI in precision medicine.

PMID:40170267 | DOI:10.1002/ctm2.70290

Afr J Prim Health Care Fam Med. 2025 Mar 31;17(1):e1-e4. doi: 10.4102/phcfm.v17i1.4929.

ABSTRACT

Prescribing cascades contribute to the increasing prevalence of polypharmacy and its associated risks, where a drug-induced adverse event is misinterpreted as a new condition and treated with additional medications. Notable cascades include the use of anticholinergics leading to cognitive impairment, dyspepsia or constipation, which then prompt prescriptions for dementia medications, proton pump inhibitors or laxatives, respectively. Similarly, calcium channel blockers and gabapentinoids often induce oedema, resulting in unnecessary diuretic use. Strategies for prevention include careful review of adverse effects, deprescribing where appropriate and clinician education to improve symptom interpretation and prescribing practices. Recognising these cascades can mitigate unnecessary interventions and improve patient outcomes.

PMID:40171689 | DOI:10.4102/phcfm.v17i1.4929

Commun Biol. 2025 Apr 2;8(1):540. doi: 10.1038/s42003-025-07944-w.

ABSTRACT

Preterm birth disrupts intestinal epithelial maturation, impairing digestive and absorptive functions. This study integrates analysis of single-cell RNA sequencing datasets, spanning fetal to adult stages, with human preterm intestinal models derived from the ileal tissue of preterm infants. We investigate the potential of extracellular vesicles (EVs) derived from human Wharton's jelly mesenchymal stem cells to promote intestinal maturation. Distinct enterocyte differentiation trajectories are identified during the transition from immature to mature stages of human intestinal development. EV treatment, particularly with the EV39 line, significantly upregulates maturation-specific gene expression related to enterocyte function. Gene set enrichment analysis reveals an enrichment of TGFβ1 signaling pathways, and proteomic analysis identifies TGFβ1 and FGF2 as key mediators of EV39's effects. These treatments enhance cell proliferation, epithelial barrier integrity, and fatty acid uptake, primarily through CFTR-dependent mechanisms-unique to human preterm models, not observed in mouse intestinal organoids. This highlights the translational potential of EV39 and CFTR activation in promoting the functional maturation of the premature human intestine.

PMID:40169914 | DOI:10.1038/s42003-025-07944-w

Basic Clin Androl. 2025 Apr 1;35(1):13. doi: 10.1186/s12610-025-00260-7.

ABSTRACT

BACKGROUND: Azoospermia, the most severe form of male infertility, is categorized into two types: non-obstructive azoospermia (NOA) and obstructive azoospermia (OA), which exhibit significant genetic heterogeneity. Azoospermia factor (AZF) deletion is a common cause of NOA, whereas congenital bilateral absence of the vas deferens (CBAVD), a severe subtype of OA, is frequently linked to cystic fibrosis transmembrane conductance regulator (CFTR) gene variants. This case report is the first to document the coexistence of a partial AZFa microdeletion and a homozygous CFTR variant in a CBAVD-affected azoospermic patient with intact spermatogenesis.

CASE PRESENTATION: A 32-year-old man presented with primary infertility and azoospermia. Clinical evaluation revealed CBAVD (normal hormone levels, low semen volume, pH 6.0, and absence of the vas deferens). Genetic analysis accidentally revealed a 384.9 kb AZFa deletion (sY84 and sY86, but not sY1064, 1182) that removed USP9Y but retained DDX3Y in the proband, his fertile brother, and his father. A homozygous CFTR variant (TG12-5T) was also detected in the proband and his brother and was inherited from heterozygous parental carriers. Microdissection testicular sperm extraction (micro-TESE) revealed intact spermatogenesis, confirmed by histology and immunofluorescence, indicating normal germ cell development.

CONCLUSION: This case expands the intricate genetic spectrum of azoospermia by illustrating the critical role of DDX3Y in the AZFa region in spermatogenesis and the variable penetrance of CFTR variant (TG12-5T) in CBAVD. These insights may refine diagnostic strategies and underscore the necessity for tailored fertility management in individuals with multifactorial genetic anomalies.

PMID:40169970 | DOI:10.1186/s12610-025-00260-7

MAbs. 2025 Dec;17(1):2483944. doi: 10.1080/19420862.2025.2483944. Epub 2025 Apr 1.

ABSTRACT

Highly concentrated antibody solutions are necessary for developing subcutaneous injections but often exhibit high viscosities, posing challenges in antibody-drug development, manufacturing, and administration. Previous computational models were only limited to a few dozen data points for training, a bottleneck for generalizability. In this study, we measured the viscosity of a panel of 229 monoclonal antibodies (mAbs) to develop predictive models for high concentration mAb screening. We developed DeepViscosity, consisting of 102 ensemble artificial neural network models to classify low-viscosity (≤20 cP) and high-viscosity (>20 cP) mAbs at 150 mg/mL, using 30 features from a sequence-based DeepSP model. Two independent test sets, comprising 16 and 38 mAbs with known experimental viscosity, were used to assess DeepViscosity's generalizability. The model exhibited an accuracy of 87.5% and 89.5% on both test sets, respectively, surpassing other predictive methods. DeepViscosity will facilitate early-stage antibody development to select low-viscosity antibodies for improved manufacturability and formulation properties, critical for subcutaneous drug delivery. The webserver-based application can be freely accessed via https://devpred.onrender.com/DeepViscosity.

PMID:40170162 | DOI:10.1080/19420862.2025.2483944

BMC Bioinformatics. 2025 Apr 1;26(1):96. doi: 10.1186/s12859-025-06092-6.

ABSTRACT

BACKGROUND: Chromatin loops are critical for the three-dimensional organization of the genome and gene regulation. Accurate identification of chromatin loops is essential for understanding the regulatory mechanisms in disease. However, current mainstream detection methods rely primarily on single-source data, such as Hi-C, which limits these methods' ability to capture the diverse features of chromatin loop structures. In contrast, multi-source data integration and deep learning approaches, though not yet widely applied, hold significant potential.

RESULTS: In this study, we developed a method called DconnLoop to integrate Hi-C, ChIP-seq, and ATAC-seq data to predict chromatin loops. This method achieves feature extraction and fusion of multi-source data by integrating residual mechanisms, directional connectivity excitation modules, and interactive feature space decoders. Finally, we apply density estimation and density clustering to the genome-wide prediction results to identify more representative loops. The code is available from https://github.com/kuikui-C/DconnLoop .

CONCLUSIONS: The results demonstrate that DconnLoop outperforms existing methods in both precision and recall. In various experiments, including Aggregate Peak Analysis and peak enrichment comparisons, DconnLoop consistently shows advantages. Extensive ablation studies and validation across different sequencing depths further confirm DconnLoop's robustness and generalizability.

PMID:40170155 | DOI:10.1186/s12859-025-06092-6

BMC Med Inform Decis Mak. 2025 Apr 1;25(1):156. doi: 10.1186/s12911-025-02983-z.

ABSTRACT

BACKGROUND: Risk stratification of COVID-19 patients can support therapeutic decisions, planning and resource allocation in the hospital. In times of high incidence, a prognostic model based on data efficiently retrieved from one source can enable fast decision support.

METHODS: A model was developed to identify patients at risk of developing severe COVID-19 within one month based on their age, sex and imaging features extracted from the thoracic computed tomography (CT). The model was trained on publicly available data from the Study of Thoracic CT in COVID-19 (STOIC) challenge and validated on unseen data from the same study and an external, multicentric dataset. The model, trained on data acquired before any variant of concern dominated, was assessed separately on data collected at later stages of the pandemic when the delta and omicron variants were most prevalent.

RESULTS: A logistic regression based on handcrafted features was found to perform on par with a direct deep learning approach, and the former was selected for simplicity. Volumetric and intensity-based features of lesions and healthy lung parenchyma proved most predictive, in addition to patient age and sex. The model reached an area under the curve of 0.78 on the challenge test set and 0.74 on the external test set. The performance did not drop for the subset acquired at a later stage of the pandemic.

CONCLUSIONS: A logistic regression utilizing features from thoracic CT and its metadata can provide rapid decision support by estimating short-term COVID-19 severity. Its stable performance underscores its potential for real-world clinical integration. By enabling rapid risk stratification using readily available imaging data, this approach can support early clinical decision-making, optimize resource allocation, and improve patient management, particularly during surges in COVID-19 cases. Furthermore, this study provides a foundation for future research on prognostic modelling in respiratory infections.

PMID:40170034 | DOI:10.1186/s12911-025-02983-z

BMC Ophthalmol. 2025 Apr 1;25(1):162. doi: 10.1186/s12886-025-03988-y.

ABSTRACT

PURPOSE: To compare two artificial intelligence (AI) models, residual neural networks ResNet-50 and ResNet-101, for screening thyroid eye disease (TED) using frontal face photographs, and to test these models under clinical conditions.

METHODS: A total of 1601 face photographs were obtained. These photographs were preprocessed by cropping to a region centered around the eyes. For the deep learning process, photographs from 643 TED patients and 643 healthy individuals were used for training the ResNet models. Additionally, 81 photographs of TED patients and 74 of normal subjects were used as the validation dataset. Finally, 80 TED cases and 80 healthy subjects comprised the test dataset. For application tests under clinical conditions, data from 25 TED patients and 25 healthy individuals were utilized to evaluate the non-inferiority of the AI models, with general ophthalmologists and fellowships as the control group.

RESULTS: In the test set verification of the ResNet-50 AI model, the area under the receiver operating characteristic (ROC) curve (AUC), accuracy, sensitivity, and specificity were 0.94, 0.88, 0.64, and 0.92, respectively. For the ResNet-101 AI model, these metrics were 0.93, 0.84, 0.76, and 0.92, respectively. In the application tests under clinical conditions, to evaluate the non-inferiority of the ResNet-50 AI model, the AUC, accuracy, sensitivity, and specificity were 0.82, 0.82, 0.88, and 0.76, respectively. For the ResNet-101 AI model, these metrics were 0.91, 0.84, 0.92, and 0.76, respectively, with no statistically significant differences between the two models for any of the metrics (all p-values > 0.05).

CONCLUSIONS: Face image-based TED screening using ResNet-50 and ResNet-101 AI models shows acceptable accuracy, sensitivity, and specificity for distinguishing TED from healthy subjects.

PMID:40169995 | DOI:10.1186/s12886-025-03988-y

BMC Oral Health. 2025 Apr 1;25(1):465. doi: 10.1186/s12903-025-05827-4.

ABSTRACT

BACKGROUND: The aim was to fully automate molar teeth developmental staging and to comprehensively analyze a wide range of deep learning models' performances for molar tooth germ detection on panoramic radiographs.

METHODS: The dataset consisted of 210 panoramic radiographies. The data were obtained from patients aged between 5 and 25 years. The stages of development of molar teeth were divided into 4 classes such as M1, M2, M3 and M4. 9 different convolutional neural network models, which were Cascade R-CNN, YOLOv3, Hybrid Task Cascade(HTC), DetectorRS, SSD, EfficientNet, NAS-FPN, Deformable DETR and Probabilistic Anchor Assignment(PAA), were used for automatic detection of these classes. Performances were evaluated by mAP for detection localization performance and confusion matrices, giving metrics of accuracy, precision, recall and F1-scores for classification part.

RESULTS: Localization performance of the models varied between 0.70 and 0.86 while average accuracy for all classes was between 0.71 and 0.82. The Deformable DETR model provided the best performance with mAP, accuracy, recall and F1-score as 0.86, 0.82, 0.86 and 0.86 respectively.

CONCLUSIONS: Molar teeth were automatically detected and categorized by modern artificial intelligence techniques. Findings demonstrated that detection and classification ability of deep learning models were promising for molar teeth development staging. Automated systems have a potential to alleviate the burden and assist dentists.

TRIAL REGISTRATION: This is retrospectively registered with the number 2023-1216 by the university ethical committee.

PMID:40169944 | DOI:10.1186/s12903-025-05827-4