Search Results (185)

Background: Obesity is a chronic, multifactorial disease that demands personalized and sustainable management approaches. Digital health technologies, such as artificial intelligence, wearable devices, mobile health apps, and virtual reality (VR), may support obesity care by providing enhanced behavioral monitoring, personalized feedback, and patient engagement. Objective: This critical narrative review discusses the current evidence on artificial intelligence, wearable technologies, and VR in the context of precision nutrition and obesity management and their possible clinical applications and limitations. Method: A critical narrative review was conducted using peer-reviewed literature published between 2019 and 2026 and identified through PubMed and Google Scholar. Search terms included combinations of “precision nutrition,” “personalized nutrition,” “obesity,” “weight management,” “metabolic health,” “digital health,” “artificial intelligence,” “machine learning,” “mobile health,” “wearable devices,” and “omics” using Boolean operators. Evidence from randomized controlled trials, systematic reviews, meta-analyses, and key conceptual studies was critically synthesized due to substantial heterogeneity in interventions and outcomes. Result: Wearables and mobile applications can enable continuous self-monitoring of physical activity, dietary intake, sleep, and physiological measures. Artificial intelligence may improve dietary personalization, risk prediction, glycemic control, and adaptive feedback. VR offers an immersive way to tackle behavioral and cognitive mechanisms related to overeating such as cravings, food cue reactivity, and inhibitory control. However, the evidence is heterogeneous, with many studies limited by short follow-up periods, small samples, variable adherence, and insufficient clinical validation. Conclusions: Artificial intelligence, wearable technologies, and VR are promising tools for precision obesity management, but their long-term clinical effectiveness remains uncertain. Future research should prioritize adequately powered trials, longer follow-up, standardized outcomes, transparent algorithms, ethical data governance, and integration with multidisciplinary nutrition and obesity care. Full article

Sexually transmitted infections (STIs) remain a major global public health challenge, while stigma, privacy concerns, and barriers to healthcare access continue to delay diagnosis and treatment. This narrative review summarizes the current evidence on the use of artificial intelligence (AI) to support STI self-assessment and digital sexual healthcare while critically discussing its current applications, challenges, and limitations, based on a PubMed literature search. Existing studies demonstrate the potential of machine-learning algorithms for individualized HIV/STI risk prediction, symptom-based assessment, automated evaluation of genital lesions, and differentiation of sexually transmitted from non-sexually transmitted conditions, with reported diagnostic performance ranging from AUCs of approximately 0.75–0.95 for symptom assessment models up to 0.893 for multimodal image-based lesion classification and validation accuracies of 94.4% for real-world image analysis platforms. Advances in computer vision and radiomics further highlight the ability of image-based AI systems to identify diagnostically relevant lesion characteristics while improving model interpretability. In parallel, generative AI chatbots are increasingly being explored as tools for sexual health education, personalized risk assessment, behavioural support, triage, and linkage to care. These technologies also influence psychological aspects of sexual health by providing private and accessible support, facilitating risk appraisal, and addressing anxiety associated with STI concerns. However, insufficient demographic diversity and the reliance on retrospective training datasets, privacy and data security concerns, limited prospective validation, and uncertainty regarding real-world clinical effectiveness remain important barriers to the widespread adoption of these technologies. To address some of these limitations, future developments are expected to focus on multimodal and explainable AI systems integrated within digital sexual health ecosystems, with clinicians remaining central to the validation, interpretation, and contextualization of AI-generated assessments and the delivery of patient-centred care. Full article

Hantaviruses, the etiological agents of hemorrhagic fever with renal syndrome (HFRS) and hantavirus pulmonary syndrome (HPS), represent a high-risk zoonotic threat with substantial global health impact. Currently, there is no FDA-approved vaccine. The viral surface glycoprotein (GP) is crucial for host cell entry and is regarded as a key target for vaccine development. However, its variability among hantavirus species limits the effectiveness of conventional vaccine strategies. Epitope-based vaccines offer a promising alternative by enabling the design of broadly protective constructs. In this study, we applied immunoinformatics approaches to design a universal multi-epitope vaccine candidate targeting both HFRS- and HPS-associated hantaviruses through a multi-layered workflow integrating B-cell and T-cell epitope prediction, antigenicity scoring, IFN-γ induction potential, conservation analysis, and population coverage assessment. Viral GPs from SEOV, PUUV, SNV, and ANDV were analyzed using algorithms for B-cell and T-cell epitope prediction. Predicted epitopes were assessed for allergenicity, toxicity, conservation, and population coverage. Two vaccine constructs incorporating β-defensin or 50S ribosomal protein L7/L12 as adjuvants were assessed for physicochemical properties, structural stability and immunogenic potential. Molecular docking analyses provided exploratory ectodomain-compatibility screening, suggesting potential interactions with TLR4 that require future experimental confirmation. The in silico immune simulations suggested potential robust and long-lasting responses with memory cell persistence exceeding one year. Simulations also indicated balanced humoral and cellular responses, robust antibody production, and long-term memory formation suggestive of durable protective immunity. These findings support the rational design of broad-spectrum multi-epitope vaccines against genetically diverse hantaviruses, offering a rational framework for preclinical development of next-generation universal vaccines against hantavirus-associated diseases. Full article

Less attention has been paid to soils with potentially toxic elements (PTEs) below agricultural land risk screening values, even though they continue to accumulate these elements. In this study, four typical areas in Ningxia were selected to determine the concentrations of cadmium (Cd) and arsenic (As) in 176 samples collected from 130 sampling sites across seven matrices. A refined mass balance model was developed by partitioning irrigation input into suspended-solid and supernatant phases and crop removal into grain and straw components. The model was used to analyze the contributions of various input and output factors and to predict future soil Cd/As concentrations and their related health risks via soil exposure. The input fluxes of Cd and As in the four regions ranged from 1.31 to 3.25 g·ha⁻¹·yr⁻¹ and 71.43 to 146.99 g·ha⁻¹·yr⁻¹, respectively, mainly contributed by irrigation water (40~64%), especially suspended solids in irrigation water, and atmospheric deposition (23~40%). The output fluxes of Cd and As were 0.89~1.43 g·ha⁻¹·yr⁻¹ and 13.81~33.86 g·ha⁻¹·yr⁻¹, respectively, dominated by crop harvesting (36~82%). The differences in input and output fluxes were mainly caused by the regional industrial structure and agricultural planting structure. The predicted results showed that soil Cd and As concentrations in all regions would not exceed regulatory limits after 100 years in the current scenario. A health risk assessment based on soil ingestion, dermal contact, and inhalation showed that the hazard indices for Cd and As were negligible, but their total carcinogenic risk reached notable levels. Over time, Cd-specific carcinogenic risk for children increased in several scenarios and transitioned from negligible to notable risk, with soil ingestion being the dominant exposure pathway. According to the results, targeted mitigation strategies, including the regulation of atmospheric deposition, optimization of irrigation water quality, and adoption of straw off-field practices, show potential to effectively limit the accumulation of potentially toxic elements in agricultural soils. Full article

Background/Objectives: Artificial intelligence (AI) is increasingly used in youth mental health services, including clinical decision support, risk prediction, and digital therapeutics. However, existing frameworks provide limited guidance for integrating ethical, cultural, and relational considerations into the design, governance, and implementation of AI-enabled mental health technologies. This scoping review examined how digital humanities-informed approaches have been incorporated into AI-supported mental health interventions for children and adolescents. Methods: A scoping review was conducted following the PRISMA Extension for Scoping Reviews (PRISMA-ScR) guidelines. Peer-reviewed literature published between 2015 and 2025 was searched using PubMed and supplemented by semantic searches through Elicit. Systematic reviews, scoping reviews, and meta-analyses examining AI-enabled digital mental health interventions and digital humanities perspectives were included. Data were synthesized using inductive thematic analysis. Results: Seventeen review-level studies met the inclusion criteria. Six recurring themes were identified: engagement, participatory co-design, human oversight, equity, ethical governance, and implementation. Across the included reviews, humanities-informed approaches were associated with greater attention to relational engagement, stakeholder participation, transparency, contextual adaptation, and culturally responsive implementation. Evidence supporting intervention effectiveness was strongest in systematic reviews and meta-analyses, whereas findings related to ethics, governance, equity, and implementation were derived primarily from scoping reviews and conceptual syntheses. Conclusions: This review suggests that digital humanities provides a valuable interdisciplinary perspective for informing the design, governance, and implementation of AI-enabled youth mental health interventions. Although the current evidence base remains heterogeneous, integrating humanities-informed approaches may support the development of AI systems that are more ethical, equitable, and developmentally responsive. Future research should evaluate these approaches through empirical implementation studies and emerging generative AI applications. Full article

The ongoing 2026 Bundibugyo ebolavirus outbreak in eastern Democratic Republic of the Congo highlights the continued vulnerability of Central Africa to recurrent Ebola emergence. This outbreak emerged less than six months after the previous one ended, appears to represent one of the shortest documented inter-epidemic intervals in the Democratic Republic of the Congo based on currently available outbreak reports. Current surveillance systems remain largely reactive, focusing on the detection of human cases after zoonotic spillover has occurred. While strengthening health systems, diagnostic capacity, and early case detection remains the cornerstone of Ebola preparedness, growing research suggests that environmental and climatic information may eventually contribute to a broader understanding of spillover risk within a One Health framework. However, current evidence remains insufficient to identify validated environmental indicators or operational thresholds capable of predicting Ebola spillover events. Recent modeling studies have demonstrated that environmental drivers of Ebola emergence remain highly context-dependent and cannot yet support operational early warning systems. This commentary argues that continued research integrating environmental monitoring, remote sensing, ecological observations, and epidemiological data may improve understanding of Ebola emergence and eventually contribute to future preparedness strategies. Rather than proposing climate-informed preparedness approaches as an operational prediction tool, we emphasize its potential as a complementary research priority requiring further validation before any operational implementation. Full article

Cardiovascular disease remains the leading global health challenge, claiming approximately 19.8 million lives annually. The convergence of wearable technology and artificial intelligence represents a transformative shift in cardiovascular healthcare, enabling continuous real-time monitoring beyond conventional clinical settings. This narrative review synthesises current evidence on integrating consumer-grade and medical-grade wearable devices with AI algorithms for continuous cardiovascular monitoring applications, with particular attention to real-world translational applicability and global health equity. This review examined the technological landscape of wearable cardiovascular monitoring devices, including smartwatches with photoplethysmography and electrocardiogram capabilities, continuous cardiac monitoring patches, and emerging biosensor technologies. Also, the review explored AI methodologies, particularly machine learning and deep learning architectures, employed in processing complex physiological data streams from these devices. Clinical applications demonstrate impressive capabilities: arrhythmia detection with sensitivity rates exceeding 98%, continuous blood pressure monitoring through cuffless technologies, heart failure decompensation prediction, and cardiovascular risk stratification. However, substantial challenges persist, including data quality assurance, algorithm interpretability, regulatory compliance, and seamless clinical workflow integration. Privacy concerns, health disparities in algorithm performance, and the need for robust validation across diverse populations remain critical considerations. AI-enhanced wearable systems hold considerable potential for shifting cardiovascular care from reactive treatment paradigms towards predictive, preventive, and precision medicine approaches. Future directions include edge computing architectures, federated learning approaches, personalised AI models, enhanced interoperability with electronic health records, and expansion to resource-limited settings, ultimately improving patient outcomes whilst reducing healthcare costs. Full article

In high-pier bridge construction, the quality and accuracy of embedded rebar placement are critical to ensuring structural safety and durability. However, conventional manual inspection methods are inefficient, subjective and pose significant safety risks in high-altitude operations. These methods are unable to comprehensively inspect all pier columns on a daily basis, and frequently result in delays in acceptance that necessitate rework. In order to address these challenges, the current study proposes a smart vision-based inspection framework for the automatic and high-precision quality assessment of rebar under long-distance imaging conditions. This approach allows quality inspectors to remotely predict and evaluate the embedment quality of rebars from a safe distance. Notably, this work introduces a novel dual-source coordinate fusion mechanism that integrates improved instance segmentation with corner detection for global-to-local precision enhancement, representing an original contribution to rebar placement inspection in complex high-pier scenarios. The framework integrates an improved YOLOv8-CD segmentation model and a corner detection algorithm through a dual-source coordinate fusion mechanism, achieving an integration of global rebar detection and local feature enhancement. The YOLOv8-CD model, when optimised, features the Convolutional Block Attention Module (CBAM) integrated into the backbone, with the objective of enhancing recognition accuracy for small targets. Additionally, a Dilation-Wise Residual (DWR) module has been inserted before the neck C2f layer for the purpose of strengthening multi-scale feature extraction. The process of perspective correction and pixel-to-actual-length conversion coefficienting is performed in order to achieve a millimetre-level measurement of the rebar spacing and diameter. Empirical validation through real high-pier construction scenes demonstrates that the proposed framework attains a detection accuracy of 98.82%, surpassing conventional YOLO-based and single-source methodologies. The experimental results demonstrate that this framework is able to detect objects at longer distances, and to maintain its performance when the target is at a greater distance than that which was used for training. The proposed approach is expected to provide an efficient, safe, and quantitative solution for intelligent bridge construction quality monitoring, offering valuable insights for the future development of smart construction and structural health inspection systems. Full article

The monitoring of environmental pollution has attracted growing attention to per- and polyfluoroalkyl substances (PFASs) due to their long persistence and resistance to natural degradation. Despite extensive research into the occurrence of PFASs in the environment, their interactions with biodegradable polymers remain understudied, highlighting a substantial knowledge gap regarding their adsorption capacity and associated toxicological implications. Therefore, the main objective of the present study was to evaluate PFAS adsorption onto biodegradable microplastics (BMPs) and compare their behavior to non- BMPs. Although biodegradable plastics (BPs) are generally considered environmentally friendly polymers, the present study indicates that they can act as vectors for PFASs as shown by LC-MS-MS analysis. Based on the current results, PFAS adsorption depends on the polymer type and PFAS chemical structure. Although BMPs are not the only carriers of PFASs, compared with various natural particles, their persistence and mobility can still influence the transport and bioavailability of PFASs in the environment. Future studies are required for assessing the environmental fate of PFASs in systems containing biodegradable polymer, in order to enhance predictive accuracy and mitigate associated risks to both the environment and human health. Full article

This review evaluates multi-omics applications in dairy research across nutrition, safety, and health. Through multi-omics integration, we reveal nutrient differences driven by species, rearing practices, and processing techniques, identify protein patterns and allergen profiles, and construct adulteration detection fingerprints and species-specific peptide markers, thereby improving the timeliness and accuracy of safety assessment. The coupling of metagenomics and metabolomics effectively predicts spoilage-related microbial risks, enabling better risk control. Furthermore, multi-omics approaches systematically elucidate the functional mechanisms of bioactive peptides (e.g., ACE-inhibitory peptides), clarify the prebiotic effects of functional oligosaccharides, and build interaction networks between dairy components and gut microbiota. The introduction of machine learning enables origin and shelf-life prediction, as well as the discovery of novel biomarkers, promoting personalized nutrition and precision fermentation strategies. However, the field is currently constrained by severe reproducibility issues arising from the absence of standardized operating procedures, excessive optimism regarding machine learning models that rarely generalize across laboratories or product matrices, and a persistent disconnect between laboratory-scale biomarker discovery and industrial implementation. Without rigorous cross-platform validation and openly shared multi-omics reference datasets, most published markers remain unfit for regulatory or industrial application. Future efforts should establish standardized workflows and expand the evidence base to drive the dairy industry toward safer, healthier, and more traceable directions. Full article

Helicobacter pylori (H. pylori) infection is one of the major causes of gastric cancer and remains an important global health concern. However, the biological processes linking chronic infection to malignant transformation are still not fully understood. Unlike previous reviews that mainly emphasized oxidative injury or individual virulence factors, this review synthesizes current evidence into an integrated redox-centered framework for gastric carcinogenesis. This framework links signaling pathways, epithelial adaptation, diagnostic interpretation, and therapeutic stratification. Current evidence indicates that persistent redox imbalance interacts with inflammatory signaling, mitochondrial dysfunction, metabolic reprogramming, epigenetic alteration, microbiome disruption, and oncogenic pathway activation throughout disease progression. Particular attention is given to the coordinated roles of NF-κB, STAT3, PI3K/AKT, HIF-1α, β-catenin, and Hippo-YAP signaling pathways. These pathways contribute to epithelial survival, chronic inflammation, genomic instability, and malignant transformation. This review additionally introduces a conceptual threshold model describing the progression from early epithelial stress to increasingly stable oncogenic reprogramming during long-standing infection. In addition, the limitations of conventional infection-centered diagnostics and non-selective antioxidant therapies are critically discussed. Emerging diagnostic approaches include oxidative injury biomarkers, transcriptomic and epigenetic profiling, artificial intelligence-assisted pathology, and multi-parameter predictive models. These approaches may improve risk stratification and facilitate earlier identification of high-risk gastric states. The translational implications further emphasize the importance of stage-specific and compartment-directed therapeutic strategies, particularly selective redox modulation and precision-guided targeting. Overall, this review provides a multidimensional perspective on H. pylori-associated gastric carcinogenesis and highlights future directions for predictive diagnostics, mechanistic stratification, and precision-based therapeutic development. Full article

Background: Cardiovascular risk prediction models are central to primary prevention strategies, yet substantial variability exists between contemporary and traditional equations used in clinical practice. In Spain, SCORE2 and REGICOR currently coexist as major cardiovascular risk assessment tools despite important methodological differences. However, evidence regarding their concordance in large occupational populations remains limited. Objective: To evaluate the agreement between SCORE2 and REGICOR in cardiovascular risk stratification among Spanish workers and to quantify the extent of cardiovascular risk reclassification associated with SCORE2 implementation. Methods: A multicenter cross-sectional study was conducted in 216,310 Spanish workers aged 40–64 years undergoing routine occupational health examinations between 2019 and 2024. Cardiovascular risk was estimated using SCORE2, REGICOR, ERICE, DORICA, Globorisk, and Framingham-based equations. High-risk categories were defined according to the thresholds recommended for each model. Agreement between categorical classifications was assessed using Cohen’s kappa coefficient, whereas Pearson correlation coefficients were calculated for continuous risk estimates. Results: SCORE2 classified 15,617 workers (7.22%) as high cardiovascular risk, whereas REGICOR identified only 4409 individuals (2.04%). Among workers classified as high risk by SCORE2, 14,387 (92.1%) were not identified as high risk by REGICOR. Agreement between SCORE2 and REGICOR was slight (kappa = 0.094), indicating minimal concordance in high-risk classification. By contrast, SCORE2 demonstrated higher agreement with Framingham hard coronary events (kappa = 0.567) and Globorisk (kappa = 0.534). Correlation analyses showed strong associations between SCORE2 and several continuous cardiovascular risk estimates, including the Framingham categorical score (r = 0.768), Framingham hard coronary events (r = 0.758), and Globorisk (r = 0.739), whereas the correlation between SCORE2 and REGICOR was substantially lower (r = 0.251). These findings indicate that strong statistical correlation does not necessarily translate into clinically meaningful agreement in cardiovascular risk categorization. Conclusions: Substantial discordance exists between SCORE2 and REGICOR in the identification of high cardiovascular risk among Spanish workers. SCORE2 consistently classified a considerably larger proportion of individuals as high risk, whereas REGICOR showed limited concordance with contemporary cardiovascular prediction models. Continued reliance on REGICOR instead of SCORE2 may lead to under-identification of workers who could currently be considered candidates for intensified primary cardiovascular prevention according to contemporary European prevention strategies. Nevertheless, the present study does not establish which model provides superior prediction of future cardiovascular events, and prospective outcome-based validation studies remain necessary. Full article

Perioperative anxiety is a common psychophysiological stress response experienced by patients before and after surgery, with a global prevalence of approximately 48%. Its occurrence is influenced by multiple factors including age, sex, type of surgery, and psychosocial determinants. The underlying pathophysiological mechanisms are complex, involving multi-system interactions such as autonomic nervous system imbalance, dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis, dysfunction of limbic system neural circuits, and neuroinflammation. Current assessment strategies are evolving from sole reliance on psychological scales toward multimodal approaches incorporating objective biomarkers including heart rate variability, cortisol, and electroencephalography. Management paradigms have shifted from traditional pharmacological premedication to integrated systems encompassing structured patient education, digital health tools, neuromodulation techniques, and cognitive behavioral therapy. However, significant gaps persist regarding standardized screening protocols, biomarker validation, and targeted intervention pathways for high-risk populations. Future management is likely to require more individualized risk assessment and intervention selection. Biomarker-based risk prediction, artificial intelligence-assisted intervention decision-making, and the deep integration of digital therapeutics such as virtual reality with existing enhanced recovery pathways will be key directions for improving patient outcomes and recovery quality. This structured narrative review summarizes current evidence on perioperative anxiety in adults, focusing on epidemiology, pathophysiological mechanisms, assessment tools, biomarkers, and multimodal management strategies. Full article

Objective: To identify demographic, clinical, and behavioural determinants of COVID-19 vaccination and antiviral uptake in Australia using the Capability, Opportunity, Motivation-Behaviour (COM-B) framework with psychometric validation and LASSO-enhanced variable selection. Methods: Cross-sectional analysis of the 2024 KAB BREATHE survey (n = 5177) of Australian adults, intentionally enriched for risk-stacked (more than 1 chronic condition). Primary outcomes included 2023/2024 COVID-19 booster receipt, future vaccine intentions, vaccine/antiviral beliefs and antiviral uptake. Predictors included demographics, chronic conditions, and domain-specific leave-one-out (LOO) COM-B scores standardised to mean = 0, SD = 1. COM-B domains were assessed using Cronbach’s alpha. Univariate and multivariable logistic regression models were complemented by LASSO penalised logistic regression with 10-fold cross-validation. Results: Among 5177 Australian adults, the mean age was 51.5 years (SD 16.5), 61.4% (3179/5177) were female, and 70.3% (3638/5177) were classified as risk-stacked. Booster uptake declined sharply from 50.8% (2023) to 19.1% (2024). Cronbach’s alpha showed poor internal consistency for Capability (α = 0.006) and Opportunity (α = −0.383) but was acceptable for full Motivation (α = 0.78). In adjusted models, age (aOR 1.02–1.03 per year), medically associated risk factors (aOR 1.66–3.51), and tertiary education (aOR 1.34–1.79) consistently predicted higher uptake and intention. Renting (aOR 0.59–0.78) and current employment (likely inversely associated with age) (aOR 0.73–0.83) were associated with lower uptake across all vaccine outcomes. Adding LOO COM-B scores substantially improved model fit (e.g., 2024 booster AUC 0.73→0.83); Motivation per SD was the strongest predictor (aOR 2.44–4.94 for vaccine outcomes, 1.52–2.49 for antivirals). LASSO models achieved CV-AUCs of 0.78–0.87. Among COVID-positive respondents (n = 2576), only 15.2% received antiviral treatment. Conclusions: Age, clinical risk, and socioeconomic factors, particularly housing tenure and employment status, are key drivers of COVID-19 preventive behaviours (either positively or negatively). The COM-B framework, when corrected for circular prediction and validated via Cronbach’s alpha and LASSO, provides substantial explanatory value. Targeted interventions should address structural barriers faced by renters and younger, employed individuals while leveraging high motivation among older adults and clinically vulnerable groups. Implications for Public Health: These findings support a shift from knowledge-based campaigns towards equity-focused, multi-level public health strategies that address structural barriers to COVID-19 vaccination and antiviral access in Australia. Full article

Sports-related injuries remain a major challenge in team sports, with important consequences for athlete health, performance, and team success. Recent advances in artificial intelligence (AI) and sensor-based monitoring technologies have enabled the integration of large volumes of training, competition, and physiological data to support injury prediction and risk modelling. However, the literature is characterised by substantial methodological diversity, limiting the ability to draw consistent conclusions. Hence, this scoping review aimed to map the existing evidence on the use of AI and sensor-based monitoring technologies for injury prediction and risk modelling in team sports, and to identify key methodological trends and research gaps. The scoping review was conducted in accordance with the PRISMA-ScR guidelines. Systematic searches were performed in PubMed and Scopus. Eligible studies included team-sport athletes and applied AI or machine learning approaches to predict injury occurrence, injury risk, or related outcomes using data derived from wearable or monitoring systems. Data were charted on study characteristics, sports and competition level, data sources, modelling techniques, validation strategies, and performance metrics. The database search yielded 123 records (PubMed: n = 37; Scopus: n = 86). After screening and eligibility assessment, 11 studies met the inclusion criteria. Most studies focused on football and rugby and relied primarily on wearable-derived data, particularly GPS and inertial sensor outputs. Common predictors included external workload variables, training exposure, previous injury history, and, in some studies, wellness or physiological markers. A wide range of models was reported, including logistic regression, decision trees, random forests, support vector machines, and neural networks. Validation strategies and reported performance varied markedly, and external validation was rarely undertaken. Across the included studies, injury risk was most consistently associated with external workload metrics, previous injury history, and internal or physiological indicators of recovery and readiness. However, current models remain limited by heterogeneous methodologies, single-team datasets, and the lack of external validation. Future research should emphasise multimodal data integration and multi-centre validation to develop reliable, interpretable, and practically applicable AI-based injury prediction systems. Full article