Search Results (4,488)

Long COVID (LC), also known as post-acute sequelae of COVID-19 infection (PASC), is a heterogeneous and debilitating chronic disease that currently affects 10 to 20 million people in the U.S. and over 420 million people globally. With no approved treatments, the long-term global health and economic impact of chronic LC remains high and growing. LC affects children, adolescents, and healthy adults and is characterized by over 200 diverse symptoms that persist for months to years after the acute COVID-19 infection is resolved. These symptoms target twelve major organ systems, causing dyspnea, vascular damage, cognitive impairments (“brain fog”), physical and mental fatigue, anxiety, and depression. This heterogeneity of LC symptoms, along with the lack of specific biomarkers and diagnostic tests, presents a significant challenge to the development of LC treatments. While several biological abnormalities have emerged as potential drivers of LC, a causative factor in a large subset of patients with LC, involves reservoirs of virus and/or viral RNA (vRNA) that persist months to years in multiple organs driving chronic inflammation, respiratory, muscular, cognitive, and cardiovascular damages, and provide continuous viral antigenic stimuli that overstimulate and exhaust CD4⁺ and CD8⁺ T cells. In this review, we (i) shed light on persisting virus and vRNA reservoirs detected, either directly (from biopsy, blood, stool, and autopsy samples) or indirectly through virus-specific B and T cell responses, in patients with LC and their association with the chronic symptomatology of LC; (ii) explore potential mechanisms of inflammation, immune evasion, and immune overstimulation in LC; (iii) review animal models of virus reservoirs in LC; (iv) discuss potential T cell immunotherapeutic strategies to reduce or eliminate persistent virus reservoirs, which would mitigate chronic inflammation and alleviate symptom severity in patients with LC. Full article

At present, there is growing interest in automated biometric identification applications. For these, it is crucial to have a system capable of accurately identifying a specific group of people while also detecting individuals who do not belong to that group. In face identification models that use Deep Learning (DL) techniques, this context is referred to as Open-Set Recognition (OSR), which is the focus of this work. This scenario presents a substantial challenge for this type of system, as it involves the need to effectively identify unknown individuals who were not part of the system’s training data. In this context, where the accuracy of this type of system is considered crucial, selecting the model to be used in each scenario becomes key. It is within this context that our work arises. Here, we present the results of a rigorous comparative analysis examining the precision of some of the most widely used models today for face identification, specifically some Convolutional Neural Network (CNN) models compared with a Vision Transformer (ViT) model. All models were pre-trained on the same large dataset and evaluated in an OSR scenario. The results show that ViT achieves the highest precision, outperforming CNN baselines and demonstrating better generalization for unknown identities. These findings support recent evidence that ViT is a promising alternative to CNN for this type of application. Full article

Artificial Intelligence (AI) is widely recognized as a force that will fundamentally transform traditional chicken farming models. It can reduce labor costs while ensuring welfare and at the same time increase output and quality. However, the breadth of AI’s contribution to chicken farming has not been systematically quantified on a large scale; few people know how far current AI has actually progressed or how it will improve chicken farming to enhance the sector’s sustainability. Therefore, taking “AI + sustainable chicken farming” as the theme, this study retrieved 254 research papers for a comprehensive descriptive analysis from the Web of Science (May 2003 to March 2025) and analyzed AI’s contribution to the sustainable in recent years. Results show that: In the welfare dimension, AI primarily targets disease surveillance, behavior monitoring, stress detection, and health scoring, enabling earlier, less-invasive interventions and more stable, longer productive lifespans. In economic dimension, tools such as automated counting, vision-based weighing, and precision feeding improve labor productivity and feed use while enhancing product quality. In the environmental dimension, AI supports odor prediction, ventilation monitoring, and control strategies that lower emissions and energy use, reducing farms’ environmental footprint. However, large-scale adoption remains constrained by the lack of open and interoperable model and data standards, the compute and reliability burden of continuous multi-sensor monitoring, the gap between AI-based detection and fully automated control, and economic hurdles such as high upfront costs, unclear long-term returns, and limited farmer acceptance, particularly in resource-constrained settings. Environmental applications are also underrepresented because research has been overly vision-centric while audio and IoT sensing receive less attention. Looking ahead, AI development should prioritize solutions that are low cost, robust, animal friendly, and transparent in their benefits so that return on investment is visible in practice, supported by open benchmarks and standards, edge-first deployment, and staged cost–benefit pilots. Technically, integrating video, audio, and environmental sensors into a perception–cognition–action loop and updating policies through online learning can enable full-process adaptive management that improves welfare, enhances resource efficiency, reduces emissions, and increases adoption across diverse production contexts. Full article

In recent years, telecom fraud remains prevalent in many regions, severely impacting people’s daily lives and causing substantial economic losses. However, previous research has mainly relied on expert knowledge for feature engineering, which lags behind and struggles to adapt to the continuously evolving patterns of fraud effectively. In addition, the extreme imbalance in fraud amounts within real communication data hinders the development of deep learning methods. In response, we propose a feature transformation method to represent users’ communication behavior as comprehensively as possible, and develop a convolutional neural network (CNN) with a Focal Loss function to identify rare fraudulent activities in highly imbalanced data. Experimental results on a real-world dataset show that, under conditions of severe class imbalance, the proposed method significantly outperforms existing approaches in two key metrics: recall (0.7850) and AUC (0.8662). Our work provides a new approach for telecommunication fraud detection, enabling the effective identification of fraudulent numbers. Full article

Cystic fibrosis (CF) newborn screening (NBS) aims to improve outcomes through early diagnosis, yet disparities in time to diagnosis remain. This study examines CFTR allele frequencies and variant panel performance among a diverse CF population in Georgia to inform recommendations for updating the NBS algorithm and improving equity. This cross-sectional study includes 969 people with CF (PwCF) from Georgia’s accredited CF centers. CFTR variant frequencies were calculated according to race and ethnicity. Panel performance was evaluated for Georgia’s current Luminex-39 variant test and three expanded panels. Statistical analyses compared detection rates across panels and demographic groups. Georgia’s diverse CF population demonstrates a unique CFTR allelic variability compared to national data. Increasing panel size enhances case identification. A panel including 719 CF-causing variants from the CFTR2 database significantly improves case detection from 93% to 97% (p = 0.002), as well as two-variant detection from 69% to 86% (p < 0.001). Detection of minoritized PwCF also improves with increasing panel size. However, even using the 719-variant panel, detection of non-Hispanic Black PwCF remains significantly lower compared to non-Hispanic White PwCF (case detection: p = 0.003; two-variant detection: p < 0.001). In conclusion, the use of expanded CFTR panels for NBS in Georgia would enhance timely diagnosis and improve equity. Full article

Millions of people worldwide suffer from asthma disease, and frequently, early diagnosis and efficient treatment are needed to enhance patient outcomes. Through an analysis of clinical and environmental characteristics, this study investigates a machine learning algorithm for predicting asthma using decision trees, K-Nearest Neighbors, random forests, and the naïve Bayes method. A dataset related to asthma disease is divided into two parts, with the first part for training consisting of around 70% and the second part for testing comprising 30%. Before dividing the subset, SMOTE is applied to balance the dataset because the dataset is unbalanced. Regarding the four algorithms, the decision tree attained better accuracy than the other algorithms. K-NN (K Nearest Neighbor) attained 97.50% accuracy, random forest attained 97.35% accuracy, naïve Bayes attained 69.99% accuracy, and the decision tree attained 67.65% accuracy. In all algorithms, the decision tree performed with high accuracy, as its prediction is 97.65% correct in detection. These algorithms can be applied to related predictive healthcare tasks. Full article

Background: Recent hypotheses suggest that mutations associated with alpha1 antitrypsin (AAT) deficiency (AATD) may influence the clinical presentation and progression of cystic fibrosis (CF). This study employs a longitudinal design to determine the prevalence of AATD mutations and assess their impact on CF. Methods: The study Finding AAT Deficiency in Obstructive Lung Diseases: Cystic Fibrosis (FADO-CF) is a retrospective cohort study evaluating people with CF from November 2020 to February 2024. On the date of inclusion, serum levels of AAT were measured and a genotyping of 14 mutations associated with AATD was performed. Historical information, including data on exacerbations, microbiological sputum isolations, and lung function, was obtained from the medical records, aiming at a temporal lag of 10 years. Results: The sample consisted of 369 people with CF (40.9% pediatrics). Of these, 58 (15.7%) cases presented at least one AATD mutation. The AATD allelic combinations identified were PI*MS in 47 (12.7%) cases, PI*MZ in 5 (1.4%) cases, PI*SS in 3 (0.8%) cases, PI*SZ in 2 (0.5%) cases, and PI*M/P_lowell in 1 (0.3%) case. The optimal cutoff value for AAT levels to detect AATD-associated mutation carriers was 129 mg/dL in the overall cohort (sensitivity of 73.0%; specificity 69.2%) and 99.5 mg/dL when excluding PI*MS cases (sensitivity 98.0%; specificity 90.9%), highlighting the need for lower thresholds in clinically severe genotypes to improve case detection. The number of mild exacerbations during the follow-up appeared to be associated with AATD mutations. Conclusions: AATD mutations are prevalent in CF and may impact certain clinical outcomes. If systematic screening was to be planned, we recommend considering the proposed cut-off points to select the population for genetic studies. Full article

Background/Objectives: Neonatal screening programmes for thyroid function testing, based on thyroid-stimulating hormone (TSH) assessment, detect both Permanent Congenital Hypothyroidism (PCH) and Transient Congenital Hypothyroidism (TCH). Maternal iodine-deficient dietary intake may result in compensatory neonatal TSH elevation; screening for Congenital Hypothyroidism (CH) is used as an indicator of the degree of iodine deficiency and of its control. In Greece, newborn screening for CH, using TSH measurement in dried blood spots (Guthrie card), began in 1979 through the Institute of Child Health (ICH). Although the general Greek population is considered iodine-replete, most pregnant Greek people are mildly iodine deficient according to the stricter WHO criteria. The aim of this retrospective study was to record the cases of TCH and the main causative factors over a 10-year period (2010–2019) in Greece, when the country was deemed to be iodine-replete. Methods: The number of births in Greece between 2010 and 2019 was retrieved from the Hellenic Statistical Authority (ELSTAT) archives: 952,109 births were recorded. The total number of newborns assessed through the ICH was 951,342 (99%). During this period, 22,391 newborns were identified to have TSH > 7 mIU/L after the second check on the initial card. Among those, 17,992 underwent retesting with a serum sample. Out of the retested newborns, 1979 were screened positive for CH and immediately began treatment with levothyroxine. We followed up with families, paediatricians, and paediatric endocrinologists to determine whether L-thyroxine therapy had been successfully discontinued for at least two months after the child’s third birthday. Successful contact was achieved with 889 individuals. From this group, 329 children had successfully discontinued thyroxine, classified as TCH. Demographic data, including gender, gestational age, and birth weight, were collected from the archives of the ICH. Maternal data, including thyroid medication use and the presence of elevated thyroid autoantibodies during pregnancy and childbirth, were also recorded. Results: Logistic regression analysis revealed that, while controlling for all other predictor variables, the odds ratio of transient hypothyroidism was 2.078 (95% CI: 1.530 to 2.821) for prematurely born children compared to those born at term. The effects of other factors on TCH versus PCH were not significant. Conclusions: It seems that prematurity is the main factor contributing to Transient Congenital Hypothyroidism in Greece, a recently iodine-replete country. Full article

In the construction industry, occupational health and safety plays a critical role in preventing occupational accidents and increasing productivity. In recent years, computer vision and artificial intelligence-based systems have made significant contributions to improving these processes through automatic detection and tracking of objects. The aim of this study was to fine-tune object detection models and integrate them with Large Language Models for (i). accurate detection of personal protective equipment (PPE) by specifically focusing on helmets and (ii). providing real-time recommendations based on the detections for supporting the use of helmets in construction sites. For achieving the first objective of the study, large YOLOv8/v11/v12 models were trained using a helmet dataset consisting of 16,867 images. The dataset was divided into two classes: “Head (No Helmet)” and “Helmet”. The model, once trained, was able to analyze an image from a construction site and detect and count the people with and without helmets. A tool with the aim of providing advice to workers in real time was developed to fulfil the second objective of the study. The developed tool provides the counts of the people based on video feeds or analyzing a series of images and provides recommendations on occupational safety (based on the detections from the video feed and images) through an OpenAI GPT-3.5-turbo Large Language Model and with a Streamlit-based GUI. The use of YOLO enables quick and accurate detections; in addition, the use of the OpenAI model API serves the exact same purpose. The combination of the YOLO model and OpenAI model API enables near-real-time responses to the user over the web. The paper elaborates on the fine tuning of the detection model with the helmet dataset and the development of the real-time advisory tool. Full article

With the rapid development of modern industry, people’s living pressures are gradually increasing, and an increasing number of individuals are affected by sleep disorders such as insomnia, hypersomnia, and sleep apnea syndrome. Many cardiovascular and psychiatric diseases are also closely related to sleep. Therefore, the early detection, accurate diagnosis, and treatment of sleep disorders an urgent research priority. Traditional manual sleep staging methods have many problems, such as being time-consuming and cumbersome, relying on expert experience, or being subjective. To address these issues, researchers have proposed multiple algorithmic strategies for sleep staging automation based on deep learning in recent years. This paper studies MASleepNet, a sleep staging neural network model that integrates multimodal deep features. This model takes multi-channel Polysomnography (PSG) signals (including EEG (Fpz-Cz, Pz-Oz), EOG, and EMG) as input and employs a multi-scale convolutional module to extract features at different time scales in parallel. It then adaptively weights and fuses the features from each modality using a channel-wise attention mechanism. The integrated temporal features are integrated into a Bidirectional Long Short-Term Memory (BiLSTM) sequence encoder, where an attention mechanism is introduced to identify key temporal segments. The final classification result is produced by the fully connected layer. The proposed model was experimentally evaluated on the Sleep-EDF dataset (consisting of two subsets, Sleep-EDF-78 and Sleep-EDF-20), achieving classification accuracies of 82.56% and 84.53% on the two subsets, respectively. These results demonstrate that deep models that integrate multimodal signals and an attention mechanism offer the possibility to enhance the efficiency of automatic sleep staging compared to cutting-edge methods. Full article

Sudden Infant Death Syndrome (SIDS) is one of the leading causes of postnatal mortality, with the prone sleeping position identified as a critical risk factor. This article presents the design, implementation, and validation of a low-cost embedded system for unobtrusive, real-time monitoring of infant posture. The system acquires data from a pressure mat on which the infant rests, converting the pressure matrix into an image representing the postural imprint. A Convolutional Neural Network (CNN) has been trained to classify these images and distinguish between prone and supine positions with high accuracy. The trained model was optimized and deployed in a data acquisition and processing system (DAQ) based on the Raspberry Pi platform, enabling local and autonomous inference. To prevent false positives, the system activates a visual and audible alarm upon detection of a sustained risk position, alongside remote notifications via the MQTT protocol. The results demonstrate that the prototype is capable of reliably and continuously identifying the infant’s posture when used by people who are not technology experts. We conclude that it is feasible to develop an autonomous, accessible, and effective monitoring system that can serve as a support tool for caregivers and as a technological basis for new strategies in SIDS prevention. Full article

The eye, in a dominant sense, can suffer disorders, such as myopia or nearsightedness, because of VDU radiation exposure. One symptom which is often caused by excessive use of VDU is eye strain. It is usually marked by an increase in the sensitivity of the eyes to light. It is known by comparing the diameter of the normal eye’s pupil and the strained eye’s pupil. People can prevent this disorder by detecting changes in the pupil’s diameter compared to the iris. Changes in the iris and pupil can be detected by using the Hough transformation to detect their shape and train perceptron neural network algorithms to recognize the patterns. As a VDI tool, an eye strain detection application can determine the condition of the user’s eyes. The level of accuracy of the method used to detect the iris and pupil using the Hough transformation is 100% for brown irises, 50% for blue irises, 33.3% for green irises, and it has a 100% accuracy in detecting an iris that is similar to the pupil and a 28.6% accuracy in detecting a pupil that is a similar color to the iris. There is also a difference in the level of accuracy of these case studies when different detection tools are used. The smartphone camera showed a 100% accuracy in detecting the iris and 28.6% accuracy in detecting the pupil. The SLR camera had a 100% accuracy in detecting the irises and 71.4% accuracy in detecting pupils, while the digital camera had 14.28% accuracy in detecting irises and a 0% accuracy in detecting a pupil. The accuracy of the perceptron algorithm in recognizing a pattern of eye strain is 70% with 20 sets of test data. Full article

Background/Objectives: The ability to drive is closely linked to participation in daily activities and quality of life in people living with neurological disorders. Cognitive deficits in people with multiple sclerosis (pwMS) are known to hinder this ability, yet concrete fitness-to-drive criteria remain elusive and assessment guidelines lack uniformity. A plethora of cognitive tests have provided associations with various aspects of driving performance and on-road behavior; however, several studies reveal limitations and inconsistencies in most tests’ sensitivity and predictive effect. Novel and resurfaced modalities for cognitive assessment, in the form of advanced imaging techniques and electrophysiological studies, may offer improved sensitivity in driving-related abilities in earlier and milder stages. Their application in addition to evaluations in driving simulators may aid future research and enhance the quality of evidence to inform decision-making. Methods: We searched for the relevant literature in the PubMed database and synthesized the available findings for the applications of currently clinically used cognitive tests, markers derived from functional magnetic resonance imaging (fMRI) and diffuse tensor imaging (DTI), as well as event-related potentials (ERP). Results: Advanced imaging modalities and ERP studies may better capture neurobiological changes that lead to driving impairment in pwMS, and they may also be applied to detect cognitive alterations earlier and with greater precision, helping to predict driving difficulties in this population. Conclusions: Novel tools and driving simulator settings could improve our understanding of the relation between cognition and driving in pwMS, enhance protocol homogeneity in driving studies, and aid in the formation of guidelines. The evidence in this review supports an increase in their application in future studies. Full article

Indoor environments significantly influence human interaction, collaboration, and well-being, yet evaluating how architectural designs actually perform in fostering social connections remains challenging. This study demonstrates the use of 11 static-mounted lidar sensors to detect serendipitous encounters—collisions—between people in a shared common space of a mixed academic–residential university building. A novel collision detection algorithm achieved 86.1% precision and detected 14,022 interactions over 115 days (67 million person-seconds) of an academic semester. While occupancy strongly predicted collision frequency overall (R² ≥ 0.74), significant spatiotemporal variations revealed the complex relationship between co-presence and social interaction. Key findings include the following: (1) collision frequency peaked early in the semester then declined by ~25% by mid-semester; (2) temporal lags between occupancy and collision peaks of 2–3 h in the afternoon indicate that social interaction differs from physical presence; (3) collisions per occupancy peaked on the weekend, with Saturday showing 52% higher rates than the weekly average; and (4) collisions clustered at key transition zones (elevator areas, stair bases), with an additional “friction effect”, where proximity to seating increased interaction rates (>30%) compared to open corridors. This methodology establishes a scalable framework for post-occupancy evaluation, enabling evidence-based assessment of design effectiveness in fostering the spontaneous interactions essential for creativity, innovation, and place-making in built environments. Full article

Coral reefs provide environmental goods and services that support biodiversity and people but face diverse threats. To assess the human factors that might be influencing the status of Cuban coral reefs, we collected and analyzed data from three sources: observations made on a research cruise that circumnavigated Cuba’s waters, expert knowledge, and updated published information. Our results show that small-scale fisheries are predominant among human factors influencing Cuban coral reefs, with more than 97% of the fishing incidents detected in situ during the expedition. Many Cuban reefs are heavily fished, have low levels of contamination, and enjoy high legal protection but experience inadequate enforcement. Tourism occurs on many reefs but could be sustainably increased based on its role in supporting enforcement and compliance and reducing fishing pressure. Densities of marine debris were generally lower in Cuban waters than other Caribbean locations and even lower within protected areas. Many human factors are likely acting synergistically, making management a challenge. This is the first at-sea comprehensive visual survey of human factors in Cuban waters and evaluation of marine debris on Cuba’s reefs, establishing a baseline for future assessments. These findings highlight potential human impacts that must be addressed to safeguard the health of Cuba’s marine ecosystem. Full article