Monitoring of advanced cancer patients' health, treatment, and supportive care is essential for improving cancer survival outcomes. Traditionally, oncology has relied on clinical metrics such as survival rates, time to disease progression, and clinician-assessed toxicities. In recent years, patient-reported outcome measures (PROMs) have provided a complementary perspective, offering insights into patients' health-related quality of life (HRQoL). However, collecting PROMs consistently requires frequent clinical assessments, creating important logistical challenges. Wearable devices combined with artificial intelligence (AI) present an innovative solution for continuous, real-time HRQoL monitoring. While deep learning models effectively capture temporal patterns in physiological data, most existing approaches are unimodal, limiting their ability to address patient heterogeneity and complexity. This study introduces a multimodal deep learning approach to estimate HRQoL in advanced cancer patients. Physiological data, such as heart rate and sleep quality collected via wearable devices, are analyzed using a hybrid model combining convolutional neural networks (CNNs) and bidirectional long short-term memory (BiLSTM) networks with an attention mechanism. The BiLSTM extracts temporal dynamics, while the attention mechanism highlights key features, and CNNs detect localized patterns. PROMs, including the Hospital Anxiety and Depression Scale (HADS) and the Integrated Palliative Care Outcome Scale (IPOS), are processed through a parallel neural network before being integrated into the physiological data pipeline. The proposed model was validated with data from 204 patients over 42 days, achieving a mean absolute percentage error (MAPE) of 0.24 in HRQoL prediction. These results demonstrate the potential of combining wearable data and PROMs to improve advanced cancer care.

Objectives This review investigates the recent advancements aimed at optimising vector control through enhanced mosquito monitoring. Recent advancements in artificial intelligence (AI), internet of things (IoT) and remote sensing have been explored to enhance surveillance but a comprehensive review of their progress was needed.Design Scoping review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines.Data sources Systematic searches were performed in PubMed and Scopus focusing on studies published between January 2014 and June 2024Eligibility criteria Original studies focusing on mosquitoes and addressing the design, development, testing, implementation, as well as the evaluation of smart traps or modification, upgrade or digitisation of traditional mosquito traps or surveillance systems were included.Data extraction and synthesis Key information relevant to this review was extracted into an Excel spreadsheet and narratively synthesised. We assessed the scientific quality of the studies included in this review based on their designs and especially on their resulting outcomes.Results This work highlights the increasing integration of AI in mosquito identification and classification, complemented by environmental sensors for real-time monitoring of parameters such as temperature, humidity and landscape features. Additionally, the convergence of AI with IoT systems, robotic traps and remote sensing technologies, including drones and geographic information systems, enables the collection of granular data on mosquito populations and their habitats, facilitating proactive control measures. Overall, these smart surveillance technologies offer a transformative advancement in vector control, enabling more precise and efficient interventions. However, their widespread adoption will require thorough evaluation of cost-effectiveness, scalability and adaptability across diverse settings.Conclusions This review maps the evolution of smart mosquito surveillance over the past decade, highlighting AI-driven species identification, IoT-enabled real-time monitoring and the integration of remote sensing technologies. It provides a structured analysis of how these innovations improve mosquito detection and data accuracy. These insights can guide further research, inform practical implementations and shape policy decisions in public health and mosquito surveillance.

Automated pupillometry is an emerging non-invasive tool for screening, diagnosis, and monitoring of ocular diseases. From biomedical engineering perspective, identifying the most useful biomarkers of the pupillary light reflex and post-illumination pupil responses is important in the automated pupillometry device design. This systematic review aims to evaluate the use of pupillometric features of the five most prevalent eye conditions: glaucoma, diabetic retinopathy, cataracts, macular degeneration, and refractive errors. Results were used to identify hardware and software requirements to specify the stimulation and analysis capabilities for early detection. Following PRISMA guidelines, a systematic search was conducted in Scopus database and studies were appraised through MMAT tool. Sixty-one studies were included, with findings synthesized by disease and stimulation protocol and interpreted by biomedical relevance and technical feasibility. Key findings suggest that glaucoma can be detected by binocular stimulation and recording to measure Relative Afferent Pupillary Defect. Diabetic retinopathy shows focal stimulation asymmetry between central and peripheral regions. In macular degeneration, focal foveal stimulation through fixation tracking and continuous analysis during phasic red and blue light stimulation can differentiate affected eyes. Refractive error studies suggest predictive potential, particularly using alternating red and blue wavelength stimuli, and comparing to corresponding time points with single colour stimuli. This review establishes a specification framework for detectable biomarkers and device capabilities required of an automated pupilometer to be most useful in early detection, including focal and full-field stimulation, binocular recording, alternating wavelength delivery, and real-time analysis.

IntroductionCommunity-based interventions represent a strategic approach to integrating the salutogenic model, involving multi-sectoral stakeholders to improve the health of specific communities. In the context of the EU's ageing population, eHealth technologies provide valuable solutions by improving older individuals' health and well-being through better access to knowledge, strengthening environmental relationships, and supporting the sustainability of health systems. This manuscript explores a community-based health promotion intervention focused on eHealth apps tested across six European regions within the GATEKEEPER project.MethodAn observational studies were conducted in the European regions of Basque Country (Spain), Aragon (Spain), Saxony (Germany), Puglia (Italy), Lodz (Poland), and Central Greece and Attica (Greece). Qualitative techniques were used to evaluate the implementation process of the community-based interventions, focused on the analysis of the facilitators, barriers, and contextual factors influencing delivery and end-users engagement to the technologies tested.ResultsSeveral factors influenced the success of the interventions, including the customisation and adaptation of applications to users' specific needs, the provision of incentives to promote engagement, and the support from health and community professionals. Customising apps to be user-friendly and culturally relevant ensures accessibility for diverse populations, while adaptation addresses varying levels of health literacy and digital skills. Continuous support from professionals fosters trust, reduces barriers to adoption, and promotes sustained engagement.DiscussionThis study provides insights into factors influencing the implementation process and adherence to digital health interventions. Understanding adherence, intention to use, and dropout rates is essential for identifying the factors contributing to the limited impact of digitally-enabled real-world interventions. The findings stress the importance of co-designing interventions, ensuring user involvement from the beginning, which improves the alignment of technology with users' needs and increases engagement.

Network analysis was used to investigate the interrelationships among 23 digital biomarkers characterizing lifestyle activity (derived from a smartwatch) and glycated haemoglobin (HbA1c) values over 18 months in 61 Type 2 diabetes patients with several comorbidities (hypertension, ischaemic heart disease, chronic kidney disease and chronic obstructive pulmonary disease) and who were taking 11 types of medications. Five lifestyle data categories—sleep, sleep stages, step daily counts, exercise, and heart rate—were analysed and included in the networks. The findings from this research not only reinforce those of previous studies but also suggest that the relationship between lifestyle and HbA1c depends not only on the type of comorbidity or medication but also on specific lifestyle behaviours. For example, frequent deep sleep is strongly correlated with a reduction in HbA1c levels on subsequent days. The evaluation of the centrality measures provides new insights into the relative importance of individual lifestyle characteristics within various networks that can be considered targets for HbA1c interventions.

The large-scale deployment of digital health solutions requires robust operational frameworks capable of coordinating heterogeneous settings, diverse stakeholders, and complex technical infrastructures. However, actionable guidance for executing federated, multinational eHealth pilots remains limited in the implementation literature. Methods: Using a mixed-methods approach, including iterative focus groups, co-creation sessions, and a Delphi study, we developed and refined a practice-derived management framework over four years within the GATEKEEPER project (EU Horizon 2020, Grant Agreement No. 857223). The study involved a federation of four European large-scale pilots. A panel of 23 experts ranked 45 best practices across six operational domains: Engagement, Intervention, Monitoring and Control, Planning, Recruitment, and Other. Results: The resulting framework integrates a structured definition of operative key performance indicators, standardised reporting and analysis tools, and a Business Intelligence dashboard to support real-time monitoring and decision-making across the preparation, deployment, and running phases of large-scale pilots. Among the ranked best practices, usability testing, user-centred digital tool design, and active recruitment strategies emerged as top priorities across pilot sites. Discussion: This management framework addresses a critical gap in implementation science by offering actionable, consensus-validated guidance for coordinating large, distributed, multi-site digital health deployments. The GATEKEEPER experience demonstrates how structured operational governance and shared performance monitoring can support the execution of complex eHealth pilots, with insights that may inform future large-scale initiatives seeking sustainable and patient-centred digital health integration.

Regulatory frameworks that integrate environmental sustainability into the lifecycle of medical devices (MDs) are essential to ensure quality, safety, and effectiveness for patients while minimizing environmental impact. The Medical Device Regulation 2017/745/EC (MDR) establishes the core framework for MDs, but additional EU legislation addresses Ecodesign, sustainable packaging, financial incentives, and waste management. Although sustainability is not explicitly included in the MDR, understanding how complementary EU regulations contribute to the European Green Deal agenda is crucial to inform decisionmakers and guide future integration of sustainability principles into medical device governance. We employed a validated policy mapping methodology, derived from the scoping review approach and adapted to systematically identify and analyse regulatory documents from policy repositories rather than academic databases. This method has been previously applied in diverse policy domains, including health, education, and digital innovation. Findings were reported according to the PRISMA-ScR. Eight binding regulations that are either directly applicable or transferable to MDs in the European Union were identified. Together, they introduce requirements on Ecodesign, packaging, financial incentives, and waste management. These complement the MDR framework by embedding sustainability principles into various stages of the MD lifecycle, even though they are not explicitly mandated within the MDR itself. While environmental sustainability provisions remain absent from the MDR, complementary EU regulations create an emerging framework that supports systemic economic transformation in line with the European Green Deal. Future research should examine enforcement and practical implementation of this framework across Member States to assess its impact on medical device regulation and environmental performance.

Background: Surgeons’ stress can significantly impact performance, leading to medical errors. Various factors contribute to stress, including a procedure’s complexity and surgeon experience. However, the field currently lacks a standardised approach to measuring stress in surgeons of different ages and experience levels. Materials and Methods: This systematic review evaluated heart rate variability (HRV) measures used in surgery to assess stress, considering surgeon age and experience. We searched PubMed, Scopus, and Web of Science following PRISMA guidelines, focusing on studies reporting HRV measurements in surgeons during surgery and comparing different surgeries, procedures, or surgeon experiences. Results: Out of 1821 reviewed studies, nine papers met the criteria, which involved 74 subjects. These studies reported various HRV measures, including heart rate, RMSSD, SDNN, pNN50, LF/HF ratio, LF, and HF. Although all time-domain features tended to show a negative response to stress, frequency-domain measures exhibited consistent patterns. However, these findings should be considered preliminary due to the small number of papers, high heterogeneity among studies, and the fact that no study has established a standard for comparing HRV across different surgeon ages or experience levels. Conclusions: Finally, these findings call for future studies with robust designs to explore the use of HRV parameters for measuring stress over time while considering surgeon age and experience.

Type 1 diabetes mellitus is a widespread chronic condition that requires insulin therapy to prevent severe complications. Effective management depends on accurate blood glucose predictions, which enable early detection of critical events. Online Learning allows models to adapt in real time by continuously incorporating new data, accommodating the dynamic nature of glycemic patterns. The increasing interest in Federated Learning for clinical use highlights the need for privacy preserving solutions capable of leveraging sensitive data from distributed sources in compliance with regulatory standards. In this work, we adapt a Federated Learning framework to the glycemic context, enabling collaborative model training across decentralized data sources for online adaptive forecasting in Type 1 Diabetes, without exposing patient information. The framework integrates three sub-models per device, each implemented as an Online Sequential Randomized Neural Network and trained on distinct curricula to specialize in specific glycemic conditions. This approach enhances predictive accuracy while minimizing computational demands. Validation on the public Ohio T1DM dataset shows that the system achieves an RMSE of 15 mg/dL with a training cost of approximately 103 FLOP per example. The proposed framework outperforms existing state of the art approaches by offering a better trade-off between accuracy and efficiency, presenting a viable alternative to larger, resource intensive architectures often unsuitable for clinical settings.

BackgroundFalls represent a major clinical and financial challenge for healthcare systems. Accurately predicting first falls remains challenging, especially when using routinely collected data.ObjectiveDevelop and evaluate a predictive model to identify elderly people at risk of first fall—defined as a fall after a 90-day fall-free period—in the Basque Country using routinely collected health records.MethodA retrospective study included patients aged ≥65 with at least two chronic conditions among heart failure, chronic obstructive pulmonary disease (COPD), and diabetes. Data on demographics, diagnoses, prescriptions and healthcare utilisation were obtained from Osakidetza-Basque Health Service databases. Patients were labelled as “fallers” if they fell during 2022–2023 after a 90-day fall-free period rather than a true first-ever fall. Predictive models—logistic regression (LR), random forest (RF), and extreme gradient boosting (XGB)—were trained using recursive feature elimination with cross-validation (RFECV). Shapley additive explanations (SHAP) enhanced model interpretability and explainability.Results35,197 patients were included, with 10.6% experiencing a fall. All models achieved similar results, with an AUCROC score of 0.71, while precision remained low. Emergency room visits in the prior 3 months, presence of caregiver, and age consistently ranked the top predictors. The number of prescriptions and antidepressant use also emerged as relevant.DiscussionModels showed moderate predictive performance. Relying solely on routinely collected health records limited clinical applicability due to low precision. Future work should integrate diverse data sources—health records, real-time gait and balance metrics, environmental factors—to improve fall prediction and support clinical decision-making.

Accurate monitoring of stress and workload in surgical environments is essential for improving performance and patient safety. Physiological signals, particularly heart rate variability (HRV), offer a promising approach for objective and continuous stress assessment. However, existing studies often rely on population-level models, limited temporal analysis, and evaluation strategies that may not reflect real-world generalisation. This study proposes an end-to-end pipeline for HRV-based classification of HRV-derived surrogate stress-like states in surgical settings, combining personalised feature extraction, surrogate label generation, and sequence-based deep learning. HRV data were collected from multiple surgeons during real colorectal procedures using wearable sensors and processed into minute-level feature representations. Stress labels were generated using subject-specific thresholds to account for individual physiological differences. Sequence-based models, including convolutional neural networks (CNNs) and long short-term memory (LSTM) networks, were evaluated across multiple temporal window lengths and validation strategies. The results show that mean-based Standard Deviation of Normal-to-Normal intervals (SDNN) provided the most consistent separation of physiological states. The highest-performing models were most frequently associated with temporal windows of 10–20 min, while CNN and LSTM architectures achieved comparable performance. Notably, in this dataset, leave-one-subject-out (LOSO) validation produced performance that was comparable to, and in some configurations higher than, pooled random splits, highlighting the influence of evaluation strategy and the potential benefits of personalised labelling. Overall, this work suggests that personalised HRV-based modelling and temporal sequence analysis may provide a feasible approach for identifying HRV-derived stress-like physiological states in real surgical environments. The findings also emphasise the importance of rigorous evaluation design when developing physiological machine learning systems.

BACKGROUND AND AIM:Hypertension is a leading cardiovascular risk factor with substantial global impact on morbidity, mortality, and healthcare costs. While lifestyle interventions remain central to management, mHealth technologies offer promising adjunctive support, though their clinical effectiveness remains uncertain. This study evaluated whether combining dietetic counseling with digital tools improves hemodynamic markers in adults aged ≥55 years with increased cardiometabolic risk. METHODS AND RESULTS:This 3-month RCT (NCT05031299) included 954 adults with at least one metabolic syndrome risk factor, allocated 1:1:1 to Standard Care (dietetic counseling), Platform (counseling plus web-based platform), or Platform + Devices (counseling plus platform plus wearables). Outcomes included anthropometrics, lifestyle characteristics, blood pressure, pulse pressure, and estimated pulse wave velocity, analyzed using linear mixed-effects models adjusted for age and sex. All groups improved over 3 months. Waist circumference decreased by -6.29, -4.92, and -4.69 cm across Standard Care, Platform, and Platform + Devices groups respectively, and systolic blood pressure declined by -4.84 to -7.15 mmHg across groups. The Platform + Devices group showed greater increases in physical activity (94.62 MET-min/week; 95% CI 66.49 to 122.76) and greater reductions in pulse pressure (-3.90 mmHg; -6.58 to -1.22) versus Standard Care. Weight loss was associated with lower odds of hypertension (OR 0.4; 95% CI 0.2-0.7), greater likelihood of hypertension reversal (OR 3.6; 1.2-10.3), and higher probability of achieving normal pulse pressure (OR 1.8; 1.1-3.1). CONCLUSIONS:Dietary lifestyle intervention improved cardiometabolic outcomes, with limited added benefit from digital tools. Weight loss was the primary driver of hemodynamic improvement.

To overcome the limitations of manual administrative coding in geriatric Cardiovascular Risk Management, this study introduces an automated classification framework leveraging unstructured Electronic Health Records (EHRs). Using a dataset of 3,482 patients, we benchmarked three distinct modeling paradigms on longitudinal Dutch clinical narratives: classical machine learning baselines, specialized deep learning architectures optimized for large-context sequences, and general-purpose generative Large Language Models (LLMs) in a zero-shot setting. Additionally, we evaluated a late fusion strategy to integrate unstructured text with structured medication embeddings and anthropometric data. Our analysis reveals that the custom Transformer architecture outperforms both traditional methods and generative llms, achieving the highest F1-scores and Matthews Correlation Coefficients. These findings underscore the critical role of specialized hierarchical attention mechanisms in capturing long-range dependencies within medical texts, presenting a robust, automated alternative to manual workflows for clinical risk stratification.

Marker-based motion capture systems are considered the gold standard for an objective assessment of threedimensional shoulder movement, but their use is limited to laboratory environments. Markerless approaches offer more flexible alternatives, yet they still require controlled settings. Thermal imaging has recently emerged as a non-invasive solution for motion analysis, while preserving user privacy, an important feature for telerehabilitation and home-based applications. Therefore, the aim of this study was to investigate the agreement between shoulder elevation angles estimated from a thermal camera with those obtained from a markerless motion capture system.Four healthy volunteers participated in this study and performed six consecutive repetitions of two bilateral functional shoulder movements of clinical relevance, namely elevation in the frontal plane up to approximately 90° and elevation in the scapular plane up to approximately 120°. Each task was performed and recorded in three separate trials. Motion data were acquired simultaneously using a SEEK Compact Pro thermal camera and a multi-camera markerless motion capture system (Qualisys).Results showed a mean absolute error between 11° and 15° between the thermal-based and markerless measurements for peak shoulder elevation angles. Linear regression analysis demonstrated strong correlations between thermal-based and markerless measurements (r = 0.89–0.98), with coefficients of determination ranging between 0.79 and 0.97 across movement planes and shoulders. These findings support the potential applicability of a single thermal camera as a markerless, affordable and accessible solution for musculoskeletal telerehabilitation approaches. Future work should focus on including a larger sample size and evaluating more complex functional tasks.

Digital innovation in the healthcare industry is transforming the delivery of healthcare services and enhancing their inherent capabilities. At the heart of this revolutionary process are healthcare data platforms, which enable organizations to aggregate and leverage patient data. However, the application of Big Data Analytics (BDA) and Artificial Intelligence (AI) in this sector presents important challenges, including technical, ethical, social, economic, organizational, and political-legal issues. The multi-disciplinary and cross-sectoral nature of the health and life-science contexts introduces barriers to communication and data sharing among various stakeholders, and exacerbates technical challenges such as interoperability, data protection, security management, compliance with laws and regulations, and support for AI applications. This paper presents the AI Big Data Platform (BDP) for healthcare, developed within the GATEKEEPER project, addressing the challenges associated with implementing state-of-the-art solutions in the healthcare context. This AI BDP facilitates the extraction of value from big volumes of heterogeneous and sensitive patient data while preserving privacy. It provides key features in interoperability, end-to-end security, multitenancy, and support for computationally intensive AI workloads. The AI Big Data Platform’s services are utilized by 8 GATEKEEPER pilots, deployed into 7 different countries, to implement 9 Reference Use Cases (RUC) 1, and involving approximately 200 users.

Robotics has been proposed as a promising solution for treating individuals with motor, sensory, and/or cognitive disabilities. Despite the great technological effort put into this field, the translation of robots from the laboratory to the clinical environment is not a seamless and smooth process, and their real-world adoption remains limited. Several barriers to the introduction of robotics in clinical practice have been identified, including a lack of sufficient scientific evidence about its actual cost/effectiveness, resistance to adopting these technologies, and economic, ethical, and regulatory restraints. Fit for Medical Robotics (Fit4MedRob) is an ambitious Initiative designed to bridge the gap between technological innovation and clinical application. One of the main goals of the Initiative is to conduct large-scale pragmatic trials to evaluate the effectiveness and the sustainability of commercially available robotic solutions. To guide the design of these trials, different online surveys have been implemented and delivered to identify the needs of healthcare practitioners and patients at different phases of the disease (acute to chronic) and therapeutic settings (hospital to home care). The results of the Initiative will suggest new organizational models to effectively introduce robotics-assisted rehabilitation into clinical practice. The paper will report on the opportunities of robotics for rehabilitation, the barriers to their clinical implementation, and the proposal of Fit4MedRob to overcome such limitations and facilitate the effective clinical implementation of robotic solutions.

AIMS:Left ventricular hypertrophy (LVH) is a common clinical finding associated with adverse cardiovascular outcomes. Once LVH is diagnosed, defining its cause has crucial clinical implications. Artificial intelligence (AI) may allow significant progress in the automated detection of LVH and its underlying causes from cardiovascular imaging. This systematic review aims to investigate the diagnostic performance of AI models developed to diagnose LVH and its common aetiologies. METHODS:MEDLINE/PubMed, EMBASE and Cochrane databases were systematically searched to identify relevant studies on echocardiography, cardiac magnetic resonance (CMR), and cardiac computed tomography (CT). RESULTS:Thirty studies were included in this review. Of them, 14 were on echocardiography, 15 on CMR, and one on cardiac CT. Regarding the AI methods applied, 79 % of studies in echocardiography utilized deep learning (DL), 64 % employed convolutional neural networks (CNNs), and 21 % applied traditional machine learning (ML) algorithms. For CMR studies, 53 % used DL, 27 % relied on CNNs, and 47 % adopted traditional ML methods. All studies showed good diagnostic performances, but those applying AI tools to determine the underlying causes of LVH demonstrated the highest accuracy metrics compared to those focused on detecting LVH itself. CONCLUSION:AI models designed to detect and differentiate LVH on cardiac imaging are currently under development and are demonstrating promising results. Further studies focusing on real-life validation of these models, and cost-effectiveness analyses are needed.

Abstract Background Oxygen therapy is critical and vital treatment for hypoxemia and respiratory distress, however, access to reliable oxygen systems remains limited in SSA. Despite WHO initiatives that distributed over 30,000 OC oxygen concentrators worldwide, SSA faces significant challenges related to their maintenance and use, due to harsh environmental conditions, technical skill shortages and inadequate infrastructure. This review aims to systematically identify and assess the literature on OC design adaptations, maintenance challenges, and knowledge gaps in SSA, providing actionable recommendations to inform innovative and context-sensitive solutions to improve healthcare delivery in the region. Methods The study focused on medical oxygen concentrators in SSA countries. It was conducted by following the PRISMA statement and searching three databases, i.e., Scopus, PubMed, and Web of Science, for publications in the period 2001–2023, using the search terms: oxygen concentrator, therapy, cylinder, plant, supply, delivery, and availability, design, and maintenance. The screening process involved evaluating manuscripts based on their titles, abstracts and full texts, based on specific inclusion and exclusion criteria. The extracted information included the author’s publication year, country, study aim, and key findings. Results Overall, 1,057 papers were returned for our analysis, of which 20 met the inclusion criteria. These studies primarily examined the design, availability and cost-effectiveness of oxygen concentrators compared to cylinders, revealing a significant supply and demand gap for these devices in SSA. It also illustrated how the environmental challenges impacted the devices durability, highlighting the need for more locally adapted resilient solutions. Solar-powered systems provide a sustainable option in areas with unstable power supplies, although initial costs remain high. Robust maintenance strategies, capacity building and strict procurement protocols proved essential to ensuring equipment long-term functionality. Conclusion This review synthesized and critically assessed the current in the body of literature, enabling highlighting valuable insights for innovators and stakeholders with an interest in enhancing the oxygen availability in SSA. It highlighted a pressing need for improved healthcare infrastructure investment, context-aware OC design and novel standards and regulatory frameworks to support frugal innovation.

This study focused on identifying emotional states by analyzing brain and physiological data, aiming to improve diagnostic and intervention strategies. Sixteen patients watched videos that elicited positive, negative, or neutral emotions, and they rated these on a 9-level valence scale. The study collected data from three sources: BOLD fMRI signals, PPG (photoplethysmography), and respiratory data across 30 trials of 25 s each. A multi-expert ensemble system was used to predict valence ratings from these physiological signals, employing advanced AI models tailored to each data type. For fMRI data, graph representations were created to capture neural activity, which were then processed using a Graph Attention Network (GAT). PPG and respiratory signals were analyzed using a sliding window approach, focusing on changes from a baseline to identify significant features. These features were processed using Fully Connected Networks (FCN). The models were trained using nested Leave-One-Subject-Out cross-validation, and their predictions were combined using a meta-learning approach, which involved training an additional AI model (either Random Forest or FCN) on the outputs of the initial models. The study demonstrated the feasibility of a multi-expert meta-learning approach for emotion detection from multi-modal physiological data, showing promise for enhancing accuracy in emotional state analysis. Further optimization of single-modality models is expected to improve performance.

Objective: This descriptive systematic review aimed to assess in the available literature on the current application and overall performance of Artificial Intelligence (AI) models in the diagnosis and classification of Rotator Cuff Tears (RCTs) using MRIs. Methods: The systematic review was performed by two of the authors from 2020 to November 2024. Only diagnostic studies involving AI application to MRI images of the rotator cuff were considered, including supraspinatus and biceps tears. Studies evaluating AI applications to Ultrasound or X-ray, or including only healthy rotator cuffs, were not analyzed in this paper. Results: The coronal plane in the T2 sequence emerged as the predominant imaging protocol, with the VGG network being the most widely utilized AI model. The studies included in this research exhibited a solid performance of the AI models with accuracy, ranging from 71.0% to 100%. The statistical analysis revealed no significant differences (p > 0.05) in accuracy, sensitivity, specificity, or precision between AI and human experts across studies that included such comparisons. Conclusions: While AI can significantly improve diagnostic efficiency and workflow optimization, future studies must focus on external validation, regulatory approval, and AI-human collaboration models to ensure safe and effective integration into orthopedic imaging.