Search Results (175)

Background: Exacerbations of chronic obstructive pulmonary disease (COPD) are a leading cause of morbidity, mortality, and healthcare burden worldwide. Early detection and timely intervention remain important challenges in COPD management, given the unpredictable nature of acute deterioration and limitations of traditional spirometry-based risk assessment. Methods: This narrative review synthesizes artificial intelligence (AI)-driven approaches for predicting and detecting chronic obstructive pulmonary disease (COPD) exacerbations across electronic health records, wearable sensors, imaging, environmental data, and patient-reported outcomes, emphasizing novel discoveries and emerging relationships rather than predictive performance. Results: Three major discoveries have been made. First, measurable physiological and behavioral deterioration may precede symptom recognition by approximately 7–14 days, thereby establishing a potential intervention window for anticipatory care. Second, machine learning (ML) models integrating pollutant exposure, medication adherence, and clinical characteristics have identified phenotypes with differential environmental sensitivity, including unexpected exposure–adherence interactions. Third, deep neural network analysis of full spirometry curves has revealed structural phenotypes beyond traditional Forced Expiratory Volume (FEV₁)-based measures and novel imaging biomarkers. The predictive performance ranges from the Area Under the Curve (AUC) 0.72–0.95, with a pooled meta-analytic AUC of approximately 0.77. Conclusions: AI has uncovered hidden patterns in the progression of COPD, supporting a shift from reactive to anticipatory management. Translation to routine care requires prospective validation, improved interpretability, workflow integration, and generalizability and equity. Full article

This systematic review examines how gait analysis has been applied to understand, detect, and manage diabetes and its complications, with a focus on wearable sensor technologies and computational methods. A total of 30 studies were identified from IEEE Xplore, Scopus, and Google Scholar databases using systematic search and screening processes. Data extraction followed a structured framework addressing research questions on gait applications, technologies, and associated parameters. Results indicate that wearable sensor technologies, coupled with advanced computational modeling and machine learning, can capture meaningful gait alterations associated with long-term metabolic dysregulation and neuropathic changes. Applications range from diabetic neuropathy detection and foot ulcer prevention to intervention evaluation and early biomarker identification. The review highlights current progress and outlines future directions toward predictive gait analytics that may serve as indirect, secondary markers of metabolic status and improve diabetes care outcomes. Furthermore, this synthesis provides evidence for integrating wearable gait assessment into diabetes management protocols, potentially enabling early detection of complications, personalized intervention strategies, and non-invasive monitoring approaches that complement traditional glucose measurements. Full article

Background/Objectives: Dental anxiety and uncooperative behavior present significant challenges in pediatric dentistry and may adversely affect treatment outcomes and oral health. The main goal of this study was to evaluate the predictive performance of machine learning models in classifying dental anxiety measured using the Abeer Children Dental Anxiety Scale (ACDAS), predicting uncooperative behavior, estimating continuous dental anxiety scores, and identifying key predictors among children aged 6–11 years attending pediatric dental clinics in Jeddah, Saudi Arabia. Methods: This is an analytical cross-sectional study conducted among 952 children to evaluate whether machine learning models could predict dental anxiety and cooperative behavior based on demographic, clinical, and behavioral variables. Twenty variables captured demographic, medical, and dental history, BMI, and anxiety/behavioral measures. Data preprocessing included removing sparse variables, imputing missing values, and encoding categorical and ordinal variables appropriately. Logistic Regression models were trained to classify dental anxiety and cooperative behavior. A Random Forest Regressor was used to predict continuous anxiety scores, and a Random Forest Classifier was used for feature importance analysis. Principal Component Analysis (PCA) and K-Means clustering were applied to explore behavioral subgroups. Results: This dataset shows the Logistic Regression model with 0.92 accuracy (ROC AUC 0.98) for predicting dental anxiety and 0.91 accuracy (ROC AUC 0.95) for cooperative behavior. The Random Forest Regressor predicted anxiety scores with R² = 0.97. Feature importance revealed that sensory and cognitive responses were key predictors of anxiety and cooperation. Unsupervised clustering identified two behavioral profiles: one with lower and another with higher anxiety and cooperation. Conclusions: ML models demonstrated strong prediction of dental anxiety and cooperation in this pediatric sample. While promising for early detection and personalized management of anxious or uncooperative children, further validation is essential before clinical use. Full article

This study examines the perceived impact of sensory and interactive architectural design in inclusive learning environments on the sensory–emotional responses and behavioral–academic outcomes of children with neurodevelopmental disorders—namely Autism Spectrum Disorder, Down Syndrome, and Attention-Deficit/Hyperactivity Disorder—during early childhood within the Egyptian educational context. Adopting a perception-based, non-causal analytical perspective, a descriptive–analytical, survey-based design was implemented using a validated questionnaire developed from an architectural–educational conceptual framework grounded in relevant literature. The study involved (N = 202) parents, teachers, therapists, and caregivers who evaluated the perceived influence of environmental design elements on children’s sensory responses, behavior, social interaction, and academic performance, based on observational and experiential assessments rather than objective environmental performance measurements. The results indicated high perceived impacts on sensory–emotional responses (84.8%) and behavioral–academic outcomes (82.0%). Movement–spatial attributes showed the strongest influence, followed by balanced natural lighting, calming colors, natural materials, and low-noise acoustic conditions, while natural elements and sensory gardens played a regulatory role in supporting emotional stability and social interaction. The study concludes that sensory- and emotionally responsive architectural design, when understood as a supportive component of the educational experience rather than an independent causal factor, and integrated with appropriate pedagogical practices, contributes to inclusive learning environments accommodating neurodevelopmental diversity, while informing the development of an applied, evidence-informed architectural design framework that translates perceptual–correlational findings into structured and operational design guidelines adaptable to the Egyptian educational context. Full article

Stress can cause many disorders, including mental and physical ones, if it persists. To take timely and effective early intervention measures, mental stress levels must be carefully monitored. This study investigates the rapidly growing topic of mental stress detection, focusing on the primary goals and mechanisms of existing detection frameworks. The main objectives and mechanisms will be highlighted. This study examines physiological sensors, stressors, algorithms, monitoring methods, and validation tools used to assess and classify mental stress. The study targets physiological sensors. Wearable sensors are becoming more popular because they can continuously monitor physiological responses in human-like environments. This allows them to reveal relevant stress patterns across various work environments. Numerous physiological sensors are used regularly. Galvanic skin response (GSR), electrocardiogram (ECG), photoplethysmography (PPG), electroencephalography (EEG), and pupil diameter camera systems are examples of these sensors. The combination of these sensors provides a wealth of cognitive and autonomic response data for stress detection. This review examines AI-based methods for interpreting complex physiological data. Machine learning and ensemble models are emphasized for improving stress classification accuracy and reducing incorrect classifications. In addition, this article discusses stressors used to induce reliable physiological responses. Validated self-report instruments are being reviewed as benchmarking tools for objective sensor-based measurements. STAI and PSS-10 are examples. These instruments demonstrate a strong correlation between stress and anxiety and physiological health outcomes. In conclusion, this review discusses future research avenues, focusing on advanced artificial intelligence-driven approaches and sophisticated sensors. These developments aim to better define stress levels and physiological factors that have not been thoroughly studied. Full article

Non-invasive monitoring technologies are increasingly being explored to support cancer care, yet most existing approaches focus on isolated parameters and fail to provide a comprehensive view of patients’ health. This study presents a feasibility evaluation of an IoT-based system designed to detect treatment-related problems in oncology patients through the integration of wearable sensors, physiological measurements, and patient-reported outcomes. A monitoring kit, including a smartwatch, tensiometer, weighing scale, and mobile device, was deployed in a cohort of 26 patients undergoing oncological treatment. Data acquisition followed a structured schedule: continuous physiological recording via the smartwatch, daily blood pressure measurements, weekly weight monitoring, and structured surveys capturing treatment-related side effects. These heterogeneous data were transformed into binary clinical metrics using rule-based feature extraction algorithms, covering conditions such as insomnia, nausea, diarrhea, abdominal pain, headache, weight loss, hypertension, and fever. Clinical specialists labeled the dataset to ensure medical validity. Machine Learning models were then trained to analyze the features and generate alerts for potential treatment complications. The results demonstrate the feasibility of integrating IoT and Artificial Intelligence techniques for continuous, patient-centered monitoring in oncology, paving the way for intelligent decision-support systems that enhance early detection and clinical management. Full article

Slow-oscillation neurofeedback (NF), encompassing slow cortical potential (SCP), infra-low-frequency (ILF), and infra-slow-fluctuation (ISF) protocols, has gained increasing interest as a non-pharmacological intervention in pediatric mental health and neurodevelopmental care. This narrative review synthesizes peer-reviewed literature on the clinical efficacy of slow-oscillation NF in children and adolescents across various conditions, including attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD), epilepsy, tic disorders, and eating-related concerns. SCP NF is the most extensively studied protocol and shows preliminary efficacy in reducing ADHD symptoms, particularly among individuals capable of learning self-regulation. For ASD and other conditions, early evidence from primarily small-scale or uncontrolled studies suggests possible benefits in emotional regulation, impulsivity, and behavioral symptoms, though findings remain mixed and often non-specific. Methodological heterogeneity, including variation in control conditions, training protocols, and outcome measures, limits the comparability of results. ILF and ISF protocols, while promising, are still emerging and require further validation. Overall, slow-oscillation NF appears to offer potential as a personalized therapeutic option for pediatric populations, but robust, well-controlled trials are needed to clarify its clinical utility and optimize its integration into multimodal care. Full article

Background: Intervention effectiveness studies rarely empirically assess Theories of Change (ToC) to determine how an intervention worked. We examine the ToC underlying the Sugira Muryango (SM) parenting program in rural Rwanda to understand whether the intervention improved child development outcomes via changes in caregivers’ behaviors to improve the home caregiving environment, as hypothesized. Methods: SM uses coaching of parents to create a safe, affectionate, stimulating, and violence-free home environment. A cluster randomized trial enrolled 1049 families with young children. SM had immediate effects on caregiver behaviors, improving scores on the Home Observation for Measurement of the Environment (HOME), harsh discipline, caregiver emotion regulation, and provision of dietary diversity. We use structural equation modeling to examine whether change in caregivers’ behaviors explains intervention-related improvements in child development (Ages and Stages Questionnaire) one year after the intervention ended. Results: Improvements in positive caregiving practices, including stimulation and early language learning as captured by the HOME, explained some of the intervention-related changes in child development, including gross motor, communication, problem-solving, and personal-social development. Increased dietary diversity explained intervention-related change in gross motor, problem-solving, and personal-social development. Change in harsh discipline and caregiver emotion regulation did not explain child outcomes. Conclusions: Intervention-related changes related to constructs captured on the HOME and dietary diversity were associated with changes in child development scores, but violent discipline and caregiver emotion regulation were not. Future research should examine whether these components of the intervention can be strengthened and may influence child development via other pathways, for example, via caregiver mental health. Full article

Open-source large language models (LLMs) are increasingly explored in educational contexts due to their transparency, adaptability, and alignment with institutional governance and equity considerations. Despite growing interest, empirical research on how open-source LLMs are deployed in education and what evidence currently supports their integration remains limited and fragmented. This paper presents a state-of-the-art narrative review of peer-reviewed, human empirical studies examining the use of open-source LLMs in education. Guided by three questions, the review synthesizes how open-source LLMs are deployed across instructional contexts, what learner-related evidence is reported, and how teachers engage in human–AI collaboration. The reviewed literature is concentrated in higher education, particularly within computer science and programming domains, with applications focused on post-class tutoring, guidance, and formative feedback. Learner perceptions are generally positive, but evidence linking open-source LLM use to measurable learning outcomes remains emerging and inconsistent. Through interpretive synthesis, the review articulates a four-role model—Designer, Facilitator, Monitor, and Evaluator—that captures how teacher agency is enacted across AI-supported instructional workflows. This review maps recurring orchestration dimensions, decision points, and tensions that characterize early implementations, and it proposes a minimal orchestration reporting scaffold (configuration, boundaries, logging, adjudication) intended to support auditability and cross-study comparison as the empirical base develops. Full article

Background: Early detection of dental caries is essential for effective oral health management. Current diagnostic workflows rely heavily on radiographic imaging, which involves infrastructure requirements, workflow coordination, and resource considerations that may limit frequent use in high-throughput or resource-constrained settings. These contextual factors motivate exploration of adjunct screening concepts that could support front-end triage decisions within existing care pathways. This study evaluates, in simulation, whether modeled tooth-percussion response signals contain sufficient discriminative information to justify further translational and managerial investigation. Implementation costs, workflow optimization, and economic outcomes are not evaluated directly; rather, the objective is to assess whether the technical preconditions for a potentially scalable screening concept are satisfied under controlled in silico conditions. Methods: An in silico model of tooth percussion was developed in which enamel, dentin, and pulp/root structures were represented as a simplified layered mechanical system. Impulse responses generated from simulated tapping were used to compute the modeled surface-vibration response (enamel-layer displacement), which served as a proxy for a measurable percussion-related signal (e.g., contact vibration), rather than a recorded acoustic waveform. Carious conditions were simulated through depth-dependent reductions in stiffness and effective mass and increases in damping to represent enamel and dentin demineralization. A synthetic dataset of labeled simulated signals was generated under varying structural parameters and measurement-noise assumptions. Machine-learning models using Mel-frequency cepstral coefficient (MFCC) features were trained to classify healthy teeth, enamel caries, and dentin caries at a screening (triage) level. Results: Under baseline simulation conditions, the classifier achieved an overall accuracy of 0.97 with balanced macro-averaged F1-score (0.97). Misclassifications occurred primarily between healthy and enamel-caries categories, whereas dentin-caries cases were most consistently identified. When measurement noise and structural variability were increased, performance declined gradually, reaching approximately 0.90 accuracy under the most challenging simulated scenario. These results indicate that discriminative information is present within the modeled signals at a screening (triage) level, meaning that higher-risk categories can be distinguished probabilistically rather than with definitive diagnostic certainty. Sensitivity and specificity trade-offs were not optimized in this study, as the objective was to assess separability rather than to define clinical decision thresholds. Conclusions: Within the constraints of the in silico model, simulated tooth-percussion response signals demonstrated discriminative patterns between healthy, enamel caries, and dentin caries categories at a screening (triage) level. These findings establish technical plausibility under controlled simulation conditions and support further investigation of percussion-based screening as a potential adjunct to clinical assessment. From a healthcare management perspective, the present results address a prerequisite question—whether such signals contain sufficient information to justify translational research, rather than demonstrating workflow optimization, cost reduction, or system-level impact. Clinical validation, threshold optimization, and implementation studies are required before managerial or operational benefits can be evaluated. Full article

Melanoma is the most lethal form of skin cancer, and early detection is critical for improving patient outcomes. Although dermoscopy combined with deep learning has advanced automated skin-lesion analysis, progress is hindered by limited access to large, well-annotated datasets and by severe class imbalance, where melanoma images are substantially underrepresented. To address these challenges, we present the first systematic benchmarking study comparing four GAN architectures—DCGAN, StyleGAN2, and two StyleGAN3 variants (T and R)—for high-resolution ($512\times 512$) melanoma-specific synthesis. We train and optimize all models on two expert-annotated benchmarks (ISIC 2018 and ISIC 2020) under unified preprocessing and hyperparameter exploration, with particular attention to R1 regularization tuning. Image quality is assessed through a multi-faceted protocol combining distribution-level metrics (FID), sample-level representativeness (FMD), qualitative dermoscopic inspection, downstream classification with a frozen EfficientNet-based melanoma detector, and independent evaluation by two board-certified dermatologists. StyleGAN2 achieves the best balance of quantitative performance and perceptual quality, attaining FID scores of 24.8 (ISIC 2018) and 7.96 (ISIC 2020) at $\gamma =0.8$. The frozen classifier recognizes 83% of StyleGAN2-generated images as melanoma, while dermatologists distinguish synthetic from real images at only 66.5% accuracy (chance = 50%), with low inter-rater agreement ($\kappa =0.17$). In a controlled augmentation experiment, adding synthetic melanoma images to address class imbalance improved melanoma detection AUC from 0.925 to 0.945 on a held-out real-image test set. These findings demonstrate that StyleGAN2-generated melanoma images preserve diagnostically relevant features and can provide a measurable benefit for mitigating class imbalance in melanoma-focused machine learning pipelines. Full article

Agile and flexible resource modelling is essential for informed decision-making in early-stage mineral project assessment, and in more advanced stages, particularly when compared with conventional deterministic geological modelling and single-estimate resource evaluations. This study presents a case of rapid scenario generation to view, interpret and test the impact of alternative geological and modelling assumptions, including the definition of geological domains, geological interpretation, grade estimation within domains, and the associated uncertainty. The workflows are implemented in Annapurna™ Resource, a cloud-native geostatistical platform designed to support agile, advanced, and multivariate modelling workflows. Focusing on the multi-commodity San Antonio–Potrerillos district, we demonstrate how rapid model construction enables the systematic evaluation of geological and statistical assumptions, contrasting deterministic estimates with probabilistic outcomes and testing their impact on estimated grades and tonnage under multiple scenarios for five elements: copper (Cu), molybdenum (Mo), gold (Au), silver (Ag), and arsenic (As). The approach provides quantitative measures of model reliability, identifies areas of high uncertainty, and supports the prioritization of new drilling to improve geological knowledge, exploration targeting, and resource classification. This case study highlights the value of fast-turnaround, probabilistic modelling not as a replacement for traditional resource reporting, but as a decision-support framework that enhances understanding of the geology, tests the sensitivity of assumptions, and accelerates learning throughout exploration and into operations. The main results suggest that additional drilling can be strategically placed to reduce the geological uncertainty derived from comparing the current interpretation with the probabilistic model built with indicator kriging. Furthermore, this has relevance in reducing the risk in the assessment of the metal content in each area of the deposit. Sensitivity analysis performed over key parameters of the estimation suggests that outliers’ treatment is the most impactful step during estimation. With current technological tools, it is possible to maintain a live resource model, which can be continuously updated to assess the impact of new data and decisions in near real time. Full article

Frequent accidents involving heavy trucks and the inadequacy of existing dynamic monitoring technologies pose significant challenges to accurate early warning risk and safety management. To address these issues, this study proposes a multi-dimensional risk measurement and real-time early warning method for heavy truck driving behavior based on trajectory data. By extracting multi-dimensional trajectory features such as lateral position, speed, and acceleration, quantitative indicators for driving stability and car-following risk were constructed. Integrated with the CRITIC objective weighting method and the K-means++ clustering algorithm, a comprehensive risk measurement model was established to systematically characterize the dynamic evolution of driving behavior, overcoming the limitations of single-dimensional risk analysis. Experimental results based on the CQSkyEyeX trajectory dataset demonstrate that the proposed method categorizes driving behavior into six risk levels. Low-risk behavior accounted for 66.70%, while medium- to high-risk behaviors mainly included serpentine driving (26.69%) and close following (4.18%). High-risk behavior constituted only 0.03%. A multi-strategy real-time warning mechanism was further developed, achieving a warning accuracy of 98.36% with the final-value method, significantly outperforming the mode method (83.62%). The outcomes of this study demonstrate the effectiveness and practical utility of the proposed model for risk identification and early warning. On a practical level, the developed risk classification framework and management strategy establish a quantitative basis for differentiated supervision, enabling a closed-loop management process of “identification–intervention–optimization”. Future work will focus on three key directions: integrating multi-source data, extending the model to other typical operational scenarios, and incorporating advanced machine learning techniques to further enhance its generalization capability and warning accuracy. Overall, this research provides a feasible technical pathway for the precise quantification, dynamic monitoring, and tiered intervention of driving behavior in heavy-duty trucks, thereby contributing to enhanced safety in road freight transportation. Full article

Machine learning models informed by patient-generated wearable data can be used to predict patient-reported and performance-based outcome measures. This approach offers a promising alternative to traditional outcome monitoring, which is commonly limited by recall bias, discrete sampling, and healthcare resource constraints. The aims of this systematic review were to identify wearable-derived features strongly associated with performance-based and patient-reported outcome measures, to compare the predictive performance across machine learning approaches, and to consolidate methodological limitations and provide suggestions for future work. Following a systematic search of four databases (PubMed, Scopus, Embase, and IEEE Xplore), 18 eligible studies were identified, published between 2017 and 2024, spanning patients across eight disease categories. Most studies used wrist-worn devices measuring accelerometry, sometimes combined with heart rate, respiratory, or sleep metrics. Random forest and support vector machine models were the most common, while hidden Markov temporal models showed improved performance with access to longitudinal data. Predictive performance ranged from poor to excellent (AUC 0.56–0.92), and non-linear models generally outperformed linear models. Despite promising early results, most studies report similar limitations of small sample sizes, limited external validation, and difficulty achieving acceptable accuracy beyond binary predictions. Overall, these studies highlight the potential of wearable-informed machine learning for continuous and objective outcome assessment, but the consensus calls for further work to apply larger, more diverse longitudinal datasets and interpretable temporal modelling approaches to bridge the gap between the current proof-of-concept state and clinical translation. Full article

Background: Targeted muscle reinnervation (TMR) is increasingly used to enhance prosthetic control and to reduce post-amputation pain. Its implementation across new centers raises questions about the reproducibility of outcomes and the impact of surgical experience. Methods: We compared the first three TMR patients treated in a newly established center in Nantes, France, with three patients treated in a high-volume center in Göttingen, Germany. Functional outcomes were measured using the Box and Block test (BBT), and operative time was recorded. Two French cases were performed with the assistance of a Göttingen-based surgeon. Conclusions: The functional outcomes showed a similar trend in both groups. The mean BBT scores were equivalent, suggesting reliable reinnervation and prosthetic integration even in early cases. Operative times were longer in Nantes, but did not impact outcomes. TMR appears not to have a pronounced learning curve, particularly regarding functional success in early cases under guided protocols. Factors such as assistance from experienced surgeons and favorable donor-to-recipient nerve ratios likely contribute to consistent outcomes. These findings support the reproducibility of TMR across institutions. Results: Within the first two years of rehabilitation we observed improvements in both functional performance and patient-reported quality of life. All six patients across both centers in-creased in BBT scores. All the patients reported an increase in social relationships and psychological health, and two of three patients reported an increase in physical health. Importantly, all six patients discontinued the use of pain medication at 2 years fol-lowing TMR. Furthermore, the French patients reported a decrease from 65–82 mm to 0–31 mm across the patients’ Visual Analog Scale (VAS) pain scores. Full article