Rui Santos

Introduction and objectives Deep learning has significantly advanced medical image analysis by enabling automated extraction of complex imaging features with high predictive performance. In forensic dental age assessment, however, most artificial intelligence studies focus on either chronological age estimation or legal age threshold classification, limiting direct comparisons between predictive tasks. This study aimed to evaluate deep learning models for both continuous age estimation and judicial age threshold classification using panoramic radiographs, assessing their performance, robustness and potential applicability as decision-support tools in forensic practice. Materials and methods A dataset of 9,814 anonymised digital panoramic radiographs from individuals with known chronological age was retrospectively collected and standardised through image preprocessing. Transfer learning applied to convolutional neural network architectures to address two complementary prediction tasks: chronological age estimation as a regression problem and legal age threshold classification at 10, 12, 14, 16, 18 and 21 years as independent binary classification problems. Model performance was evaluated using mean absolute error (MAE), coefficient of determination (R²), median absolute error and mean absolute percentage error (MAPE) for regression, and accuracy, precision, recall and F1-score for classification. Results The regression model achieved an MAE of 1.122 years, an R² of 0.917 and a MAPE of 7.67% on the independent test dataset, demonstrating good predictive performance across the analysed age range. Classification models achieved accuracy exceeding 90% with F1-scores above 0.90 for the 10-, 12-, 14- and 16-year legal thresholds. Lower predictive performance at the 18- and 21-year thresholds reflected increased biological variability and class imbalance associated with later stages of dental development. Conclusions Deep learning provides good predictive performance from panoramic radiographs for both continuous age estimation and legal age threshold classification. The complementary evaluation of regression and classification demonstrates the versatility of convolutional neural networks for forensic dental image analysis and supports their integration into trustworthy artificial intelligence systems for clinical and judicial decision support.

1. Let X1,…,Xn be a sequence of independent and identically distributed (i.i.d.) absolutely continuous random variables (r.v.) with density function f, distribution function (d.f.) F and survival f...

Age assessment is crucial in forensic and judicial decision-making, particularly in cases involving undocumented individuals and unaccompanied minors, where legal thresholds determine access to protection, healthcare, and judicial procedures. Dental age assessment is widely recognized as one of the most reliable biological approaches for adolescents and young adults, but current practices are challenged by methodological heterogeneity, fragmented data representation, and limited interoperability between clinical, forensic, and legal information systems. These limitations hinder transparency and reproducibility, amplified by the increasing adoption of AI-based methods. The AIdentifyAGE ontology is domain-specific and provides a standardized, semantically coherent framework, encompassing both manual and AI-assisted forensic dental age assessment workflows, and enabling traceable linkage between observations, methods, reference data, and reported outcomes. It models the complete medico-legal workflow, integrating judicial context, individual-level information, forensic examination data, dental developmental assessment methods, radiographic imaging, statistical reference studies, and AI-based estimation methods. It is being developed together with domain experts, and it builds on upper and established biomedical, dental, and machine learning ontologies, ensuring interoperability, extensibility, and compliance with FAIR principles. The AIdentifyAGE ontology is a fundamental step to enhance consistency, transparency, and explainability, establishing a robust foundation for ontology-driven decision support systems in medico-legal and judicial contexts.

Background: Chronological age estimation based on biological markers plays a crucial role in forensic, legal, and clinical contexts. Among available methods, dental age assessment (DAA) using orthopantomograms (OPGs) is considered one of the most accurate approaches. However, conventional DAA methods rely on expert interpretation, introducing subjectivity and limiting reproducibility. In recent years, artificial intelligence (AI) has emerged as a promising solution to overcome these limitations. This systematic review with meta-analysis aimed to evaluate the performance of AI-based models for dental age assessment using OPGs. Methods: The review protocol followed PRISMA guidelines and was registered in PROSPERO. A total of 27 studies, published between 2020 and 2025, were included, encompassing a wide range of machine learning and deep learning approaches, predominantly convolutional neural networks (CNNs). Results: For chronological age estimation, the pooled root mean squared error (RMSE) was 1.76 years, indicating that AI models can predict age with an average deviation of less than two years from true chronological age. The pooled coefficient of determination (R²) was 0.92, reflecting strong agreement and high explanatory power between predicted and actual ages. In classification tasks related to legal age thresholds, AI models demonstrated high discriminatory performance, with an overall area under the curve (AUC) exceeding 0.90 and a high diagnostic odds ratio (DOR). Hierarchical summary receiver operator characteristic analyses (HSROC) revealed high pooled sensitivity and specificity, with a slightly stronger ability to correctly classify individuals below the legal threshold. Network meta-analyses consistently ranked human expert assessments lower than AI-based models, especially CNN architectures such as EfficientNet, ResNet, and VGG16, although differences were not statistically significant. Conclusions: Despite substantial methodological heterogeneity across studies, the findings suggest that AI-based models for DAA using OPGs have reached a level of accuracy and robustness suitable for forensic and legal applications. AI systems should be regarded as complementary tools, enhancing objectivity, reproducibility, and consistency in age estimation, being good decision-support tools in legal and forensic settings.

The aim of this study was to evaluate epidemiological data on pemphigus vulgaris and pemphigoid, with special focus on characteristics, clinical manifestations, and treatment approaches. This study includes patients who have been diagnosed with pemphigus vulgaris or pemphigoid at the Dermatology Service of the Santa Maria Local Health Unit between the years 2013 and 2022. The statistical software IBM SPSS® version 29, with Clopper-Pearson 95% confidence intervals (CI) was used for proportions. The total number of patients included in this study is 99 patients with pemphigus vulgaris (49 females, 50 males) and 227 with pemphigoid patients (108 females, 119 males). Patients who have been treated for pemphigus vulgaris, the most prescribed medicine were corticosteroids, followed by the second most prescribed, drugs which were immunosuppressants. Additionally rituximab was used in 21 patients. The interval of the disease was 304 days; with an interval of no symptom disease was 415 days for relapsed patients. Similarly, for pemphigoid, the primary treatment used was with corticosteroids, which were also used for the main treatment, in many cases it was combined with other medicines, such as: antihistamines, antibiotics, and immunosuppressants. The average cure time was 148 days, with 32.0% of patients claiming they experienced relapse. To conclude, we observed that corticosteroids, particularly prednisolone, remain still the most important pharmacological treatment for pemphigus vulgaris and pemphigoid. However, rituximab is an effective pharmacological alternative for pemphigus, while the use of clobetasol propionate is increasing and becoming as one of the preferred treatments for pemphigoid.

Sir Ronald Fisher, considering it important to consistently obtain significant p-values to support an experimental hypothesis, and thus replicating experiments to yield independent p-values, was one of the first to carry out a meta-analysis of p-values to obtain an overall decision. Suspecting also that Mendel only reported the results that closely agreed with the hereditary laws that Mendel wanted to establish, Fisher denounced the research malpractice, eventually scientific fraud, of researchers replicating experiments with the intent of reporting the best results, in his words “too good to be true”, to support their expectations or claims. On the other hand, publication bias, which largely results from published studies showing evidence of an effect, is a source of temptation to repeat experiments, especially when the first one carried out does not provide a significant p-value that favors its publication. Therefore, if the minimum of two independent “genuine” Uniform(0,1) p-values is reported to increase the chances of publication, it is a “fake” Beta(1,2) p-value, and its use in a meta-analysis changes the distributions of combined tests for genuine p-values. Mendel random variables, useful for modeling mixtures of genuine and fake p-values, are studied in detail, as well as their use in testing independence versus correlation of p-values. In addition, an example shows the usefulness of tables with critical values that consider the existence of a small proportion of fake p-values, to reach an overall informed decision when carrying out a meta-analysis of independent p-values with commonly used combined tests.

Purpose: The healthcare sector is a primary target for cybercriminals, with health data breaches ranking among the most critical threats. Despite stringent penalties imposed by the U.S. Department of Health and Human Services Office for Civil Rights (OCR), vulnerabilities still persist due to slow detection and ineffective data protection measures. On the other hand, as organizations are often reluctant to disclose security breaches for fear of reputational and market share losses, penalties can serve as a useful proxy for quantifying losses and insurance claims. Methods: This study analyzes fines and settlements (2008–2024) using the traditional lognormal, general extreme value (GEV) and other heavy-tailed statistical models, including the geo-max-stable loglogistic law, and also the mixture models hyperexponential and hyperloglogistic. Results: Mixture models, either the hyperexponential or the hyperloglogistic, deliver the best fit for OCR penalties, and for yearly maxima, the best fit is achieved with the GEV distribution. Regarding Attorneys General fines, the hyperexponential model is optimal, with the GEV model excelling again for their yearly maxima. Hence, mixture models effectively capture the dual nature of penalty data, comprising clusters of moderate and extreme values. However, yearly maxima align better with the GEV model. Conclusions: The findings suggest that while Panjer’s theory for aggregate claims suffices for moderate claims, it must be supplemented with strategies to address extreme cybercrime scenarios, ensuring insurers and reinsurers can manage severe losses effectively.

Introduction Forensic Odontology plays a crucial role in medicolegal identification by comparing dental evidence in antemortem (AM) and postmortem (PM) dental records, including orthopantomograms (OPGs). Due to the complexity and time-consuming nature of this process, imaging analysis optimization is an urgent matter. Convolutional neural networks (CNN) are promising artificial intelligence (AI) structures in Forensic Odontology for their efficiency and detail in image analysis, making them a valuable tool in medicolegal identification. Therefore, this study focused on the development of a CNN algorithm capable of comparing AM and PM dental evidence in OPGs for the medicolegal identification of unknown cadavers. Materials and methods The present study included a total sample of 1235 OPGs from 1050 patients from the Stomatology Department of Unidade Local de Saúde Santa Maria, aged 16 to 30 years. Two algorithms were developed, one for age classification and another for positive identification, based on the pre-trained model VGG16, and performance was evaluated through predictive metrics and heatmaps. Results Both developed models achieved a final accuracy of 85%, reflecting high overall performance. The age classification model performed better at classifying OPGs from individuals aged between 16 and 23 years, while the positive identification model was significantly better at identifying pairs of OPGs from different individuals. Conclusions The developed AI model is useful in the medicolegal identification of unknown cadavers, with advantage in mass disaster victim identification context, by comparing AM and PM dental evidence in OPGs of individuals aged 16 to 30 years.