The rapid integration of Artificial Intelligence (AI) into healthcare promises significant benefits but also raises unprecedented ethical, clinical, and legal challenges. Current medico-legal frameworks, primarily designed for human decision-making, are often inadequate to address liability issues arising from algorithmic errors or opaque "black box" models. This paper introduces a novel medico-legal methodology that combines proactive and reactive approaches to risk assessment, originally developed within European forensic medicine, and adapts it to the context of AI in healthcare. By systematically analyzing data collection, dataset validation, error identification, and causal reconstruction, the proposed framework provides a structured path for evaluating medical liability when AI systems are involved. This dual approach not only supports clinicians, developers, and policymakers in preventing harm, but also establishes a robust forensic tool for liability assessment. The methodology offers a step toward internationally applicable standards for addressing the medico-legal implications of AI in medicine.

Forensic age assessment plays a crucial role in medico-legal contexts where an individual's chronological age cannot be reliably documented, such as immigration procedures. In recent years, artificial intelligence (AI) has increasingly been applied to automate and improve the accuracy of age estimation methods. However, the forensic applicability of these approaches, particularly in relation to legally relevant age thresholds, remains incompletely characterized. A systematic review was conducted in accordance with the PRISMA 2020 guidelines. MEDLINE (via PubMed) and Scopus were searched from database inception to 1 March 2026. Studies were included if they applied AI techniques to age estimation in a forensic or medico-legal context and reported quantitative performance metrics. The search identified 1,197 records, of which 48 studies met the inclusion criteria. Most studies investigated imaging-based approaches, particularly panoramic dental radiographs, while others explored skeletal imaging, DNA methylation markers, or emerging molecular biomarkers. Convolutional neural networks were the most frequently used modelling approach. AI models applied to dental radiographs commonly reported mean absolute errors between approximately 0.5 and 1.5 years, whereas DNA methylation-based models showed errors between 3 and 5 years. MRI-based skeletal models often achieved prediction errors close to one year. However, relatively few studies specifically evaluated classification performance around legally relevant thresholds such as 18 years. AI shows considerable potential for improving the automation and reproducibility of forensic age estimation. Further research with larger and more diverse datasets, external validation, and standardized reporting is needed to ensure reliable implementation in medico-legal practice.

Artificial intelligence (AI) has emerged as a promising tool in forensic sciences, offering new opportunities for personal identification through automated analysis of biological and imaging data. AI-based approaches have been increasingly applied to tasks such as sex estimation, human identification, ancestry estimation, and kinship analysis. This systematic review aims to synthesize the available evidence regarding the applications, methodological characteristics, and performance of AI models in forensic personal identification. A systematic literature search was conducted in PubMed/MEDLINE and Scopus following PRISMA guidelines. Studies investigating AI applications for forensic identification were included. Data extraction focused on study characteristics, dataset type, AI model architecture, forensic task, validation strategy, and reported performance metrics. A total of 89 studies published between 2012 and 2026 met the inclusion criteria. The majority of studies focused on sex estimation (63%), followed by human identification, ancestry estimation, multi-task prediction, and kinship verification. Most studies relied on imaging datasets, particularly computed tomography and radiographic images. Deep learning models represented the most frequently used analytical approaches. Reported accuracy values were generally high, with a median accuracy of 91.4% and an interquartile range of 88.9–95.0% in studies reporting single-value accuracy metrics. Deep learning approaches tended to achieve slightly higher performance than traditional machine learning models. AI shows considerable potential to support forensic personal identification, particularly in imaging-based applications. However, methodological heterogeneity, population-specific datasets, and limited external validation remain important challenges. Future research should prioritize standardized validation protocols, multi-population datasets, and transparent reporting to ensure the forensic applicability of AI-based identification systems.

Postmortem interval (PMI) estimation remains a major challenge in forensic medicine due to the complex and multifactorial nature of decomposition processes. In recent years, artificial intelligence (AI) and machine learning techniques have been increasingly applied to improve PMI prediction. This systematic review aimed to evaluate the current evidence on AI-based models developed for PMI estimation. A systematic literature search was conducted in PubMed/MEDLINE and Scopus from database inception to 1 March 2026, following PRISMA 2020 guidelines. Studies were included if they applied AI, machine learning, or deep learning methods to estimate PMI using real postmortem datasets. Data extraction included study characteristics, data modality, AI model architecture, validation strategy, and reported performance metrics. A total of 64 studies met the inclusion criteria. The most common approach involved microbiome-based models ( n = 29), followed by metabolomics and proteomics approaches ( n = 11) and imaging-based AI models ( n = 11). Random Forest algorithms were the most frequently used machine learning method, particularly in microbiome studies. Reported predictive performance varied widely across studies, with several models achieving high accuracy or low prediction errors depending on the data modality and PMI range investigated. AI represents a promising tool for improving the accuracy and objectivity of PMI estimation by enabling the integration of complex forensic datasets. However, current evidence is limited by heterogeneous methodologies, small datasets, and a lack of external validation. Future research should focus on large multicenter datasets, standardized validation protocols, and multimodal AI models integrating diverse forensic data sources.

Commotio cordis (CC) is a rare cause of sudden cardiac death resulting from a blunt, non-penetrating impact to the precordium, in the absence of underlying cardiac pathology or structural damage. Although most frequently described in adolescents engaged in sports activities, CC may also occur in non-sport contexts. We report the case of a 43-year-old male construction worker who died after being struck in the chest by a slow-moving truck. Postmortem computed tomography (PMCT) revealed multiple left rib fractures, costal cartilage fractures, manubriosternal dislocation and pulmonary contusions. Conventional autopsy confirmed the absence of cardiac rupture, coronary disease or pericardial injuries. Histological analysis supported a diagnosis of electromechanical dissociation by fatal arrhythmia. In our review of six previously published articles involving similar cases, autopsy was the principal diagnostic tool in all instances. PMCT was employed in only one case, prompted by circumstantial suspicion of child abuse. This case underscores the importance of a multidisciplinary forensic approach, combining PMCT and traditional autopsy to reconstruct the traumatic event and rule out other mechanisms of death. CC remains a diagnosis of exclusion, requiring demonstration of precordial impact, absence of lethal structural injuries, and lack of alternative causes of death on autopsy and toxicological examination. PMCT should be strongly recommended in cases of sudden death following blunt thoracic trauma, even when external signs are minimal or absent.

In the estimation of the Post-Mortem Interval (PMI), semi-quantitative methods have been proposed to overcome the challenges associated with determining the time of death. Among these, the Total Decomposition Score (TDS) method, developed by Gelderman et al., offers a systematic and semi-quantitative approach for estimating PMI. The aim of this study was to evaluate the reliability of the TDS by assessing its interoperator variability and comparing the results obtained with known reference data. A TDS-based questionnaire was administered to 100 participants - including forensic pathologists, residents in forensic medicine and professionals in forensic thanatology - using a dataset of six outdoor cadavers representing different decomposition stages. Data were analyzed using Fleiss’ Kappa (K) to assess inter-rater agreement and Spearman’s rank correlation to evaluate consistency. The results showed moderate overall agreement, with inter-rater reliability decreasing in cases with PMI exceeding 30 days. Linear regression analyses between estimated and actual post-mortem intervals yielded low coefficients of determination, with R² = 34.1% for the TDS-based model and R² = 20.5% for the ADD-based model, indicating that both methods explain only a limited portion of the variance in the actual PMI (PMIa). No statistically significant differences were observed among the professional categories, supporting the method’s applicability across different levels of expertise. While TDS shows promise as a practical tool for PMI estimation in field conditions, inter-operator variability remains a limiting factor in advanced decomposition stages.