Search Results (5,708)

Stroke is a major global cause of death and disability and necessitates both quick diagnosis and treatment within narrow windows of opportunity. CT scanning is still the first-line imaging in the acute phase, but correct interpretation may not always be readily available and may not be resource-available in poor and rural health systems. Automated stroke classification systems can offer useful diagnostic assistance, but clinical application demands high accuracy and explainable decision-making to maintain physician trust and patient safety. In this paper, a ResNet-18 model was trained on 6653 CT brain scans (hemorrhagic stroke, ischemia, normal) with two-phase fine-tuning and transfer learning, XRAI explainability analysis, and web-based clinical decision support system integration. The model performed with 95% test accuracy with good performance across all classes. This system has great potential for emergency rooms and resource-poor environments, offering quick stroke evaluation when specialists are not available, particularly by rapidly excluding hemorrhagic stroke and assisting in the identification of ischemic stroke, which are critical steps in considering tissue plasminogen activator (tPA) administration within therapeutic windows in eligible patients. The combination of classification, explainability, and clinical interface offers a complete framework for medical AI implementation. Full article

Parkinson’s disease (PD) is a progressive neurodegenerative disorder that severely impacts motor function, necessitating early detection for effective management. However, current diagnostic methods are expensive and resource-intensive, limiting their accessibility. This study proposes a non-invasive, gait-based screening approach for PD using time-series analysis of video-derived motion data. Gait patterns indicative of PD are analyzed using videos containing walking sequences of PD subjects. The video data are processed via computer vision and human pose estimation techniques to extract key body points. Classification is performed using K-Nearest Neighbors (KNN) and Long Short-Term Memory (LSTM) networks in conjunction with time-series techniques, including Dynamic Time Warping (DTW), Bag of Patterns (BoP), and Symbolic Aggregate Approximation (SAX). KNN classifies based on similarity measures derived from these methods, while LSTM captures complex temporal dependencies. Additionally, Shapelet-based Classification is independently explored for its ability to serve as a self-contained classifier by extracting discriminative motion patterns. On a self-collected dataset (43 instances: 8 PD and 35 healthy), DTW-based classification achieved 88.89% accuracy for both KNN and LSTM. On an external dataset (294 instances: 150 healthy and 144 PD with varying severity), KNN and LSTM achieved 71.19% and 57.63% accuracy, respectively. The proposed approach enhances PD detection through a cost-effective, non-invasive methodology, supporting early diagnosis and disease monitoring. By integrating machine learning with clinical insights, this study demonstrates the potential of AI-driven solutions in advancing PD screening and management. Full article

Electrocardiogram (ECG) and phonocardiogram (PCG) signals are widely used in the early prevention and diagnosis of cardiovascular diseases (CVDs) due to their ability to accurately reflect cardiac conditions from different physiological perspectives and their ease of acquisition. Currently, some studies have explored the joint use of ECG and PCG signals for disease screening, but few studies have considered the trade-off between classification performance and energy consumption in model design. In this study, we propose a multimodal CVDs detection framework based on Spiking Neural Networks (SNNs), which integrates ECG and PCG signals. A differential fusion strategy at the signal level is employed to generate a fused EPCG signal, from which time–frequency features are extracted using the Adaptive Superlets Transform (ASLT). Two separate Spiking Convolutional Neural Network (SCNN) models are then trained on the ECG and EPCG signals, respectively. A confidence-based dynamic decision-level (CDD) fusion strategy is subsequently employed to perform the final classification. The proposed method is validated on the PhysioNet/CinC Challenge 2016 dataset, achieving an accuracy of 89.74%, an AUC of 89.08%, and an energy consumption of 209.6 $\mathsf{\mu }$J. This method not only achieves better balancing performance compared to unimodal signals but also realizes an effective balance between model energy consumption and classification effect, which provides an effective idea for the development of low-power, multimodal medical diagnostic systems. Full article

Background/Objectives: Neurological disorders (ND) are a global health challenge, affecting millions and greatly reducing quality of life. Disorders such as Alzheimer’s disease, mild cognitive impairment (MCI), schizophrenia, and depression often share overlapping symptoms, complicating diagnosis and treatment. Early detection is crucial for timely intervention; however, traditional diagnostic methods rely on subjective assessments and costly imaging, which are not universally accessible. Addressing these challenges, this study investigates the classification of multiple ND using electroencephalography (EEG) signals. Methods: Various feature extraction methods were employed, and the Least Absolute Shrinkage and Selection Operator (Lasso) algorithm was utilized for effective feature selection. Two-class (disease–disease and healthy control–disease), three-class (healthy control and two ND, as well as three ND), and four-class (healthy control and three ND) classifications were conducted using different machine learning algorithms with the selected features. An EEG dataset comprising 40 Alzheimer’s patients, 43 healthy controls, 42 schizophrenia patients, 28 MCI patients, and 28 depression patients served as the experimental benchmark. Results: The Linear Discriminant Analysis (LDA) classifier achieved the highest accuracy, distinguishing between healthy controls and Alzheimer’s with 100% accuracy and demonstrating strong performance in other comparisons. Multi-class classification reached 84.67% accuracy for distinguishing depression, MCI, and schizophrenia, while four-class classification achieved 57.89%, highlighting the complexity of differentiating among multiple ND. The frequent selection of frontal lobe channels across ND indicates their critical role in classification. Conclusions: This study contributes to the literature by emphasizing disease-to-disease classification over the traditional control-versus-patient framework, highlighting the potential for more effective diagnostic tools in clinical settings. Full article

Background: Diabesity, the coexistence of obesity and type 2 diabetes, is a major public health concern. Shift work and unhealthy lifestyle behaviors may exacerbate its prevalence, particularly in working populations. Objective: This study aims to evaluate the association between sociodemographic characteristics, health behaviors, and shift work and the prevalence of diabesity, using both BMI and the CUN-BAE estimator, in a large cohort of Spanish workers. Methods: This cross-sectional study included 53,053 workers (59.8% men) aged 18–69 years who underwent occupational health examinations. Diabesity was defined as obesity (BMI ≥ 30 kg/m² or high CUN-BAE) plus fasting glucose ≥ 100 mg/dL or prior diagnosis of diabetes. Adherence to the Mediterranean diet was assessed by the MEDAS questionnaire, physical activity by the IPAQ, alcohol intake by standard drink units (UBEs), and socioeconomic class by the CNAE-11 classification. Shift work was defined according to ILO criteria. Logistic regression was used to assess associations, adjusting for potential confounders. Results: Shift work was independently associated with increased odds of diabesity both in men and women. Diabesity prevalence was higher when assessed by CUN-BAE compared with BMI. Age, male sex, lower socioeconomic class, physical inactivity, smoking, poor diet adherence, and alcohol intake were all significantly associated with higher risk. The CUN-BAE index showed superior sensitivity in identifying individuals at risk. Conclusions: Shift work and unhealthy behaviors are key determinants of diabesity among Spanish workers. The use of adiposity estimators beyond BMI, such as CUN-BAE, should be encouraged in occupational health surveillance. Workplace-targeted interventions are urgently needed to address this growing metabolic burden. Full article

Fault diagnosis under variable operating conditions remains challenging due to the limited adaptability of traditional methods. This paper proposes a transfer learning-based approach for bearing fault diagnosis across different rotational speeds, addressing the critical need for reliable detection in changing industrial environments. The method trains a diagnostic model on labeled source-domain data and transfers them to unlabeled target domains through a two-stage adaptation strategy. First, only the source-domain data are labeled to reflect real-world scenarios where target-domain labels are unavailable. The model architecture combines a convolutional neural network (CNN) for feature extraction with a self-attention mechanism for classification. During source-domain training, the feature extractor parameters are frozen to focus on classifier optimization. When transferring to target domains, the classifier parameters are frozen instead, allowing the feature extractor to adapt to new speed conditions. Experimental validation on the Case Western Reserve University bearing dataset (CWRU), Jiangnan University bearing dataset (JNU), and Southeast University gear and bearing dataset (SEU) demonstrates the method’s effectiveness, achieving accuracies of 99.95%, 99.99%, and 100%, respectively. The proposed method achieves significant model size reduction compared to conventional TL approaches (e.g., DANN and CDAN), with reductions of up to 91.97% and 64%, respectively. Furthermore, we observed a maximum reduction of 61.86% in FLOPs consumption. The results show significant improvement over conventional approaches in maintaining diagnostic performance across varying operational conditions. This study provides a practical solution for industrial applications where equipment operates under non-stationary speeds, offering both computational efficiency and reliable fault detection capabilities. Full article

Background/Objectives: Food insecurity (FI) is a well-defined factor in pediatric health outcomes and has been associated with lower diet quality. While poor diet quality has been linked to the rising prevalence of inflammatory bowel disease (IBD), little is known about the impact of FI on pediatric IBD. This pilot study explores the feasibility and potential impact of FI on dietary intake and clinical outcomes in children with newly diagnosed IBD. Methods: This pilot study included newly diagnosed IBD patients aged 5 to 18. FI screening was completed using the USDA 6-item and AAP 2-item screeners at diagnosis and 6 months. Dietary intake was classified according to their degree of processing (NOVA classification). Clinical data, anthropometrics, and healthcare utilization were collected over 6 months. Results: Among 20 patients, FI was identified in 40% of families. Food-insecure patients had significantly lower weight and BMI z-scores at diagnosis compared to food-secure peers (p = 0.002 and p = 0.0013, respectively). Food-insecure patients consumed more ultra-processed foods (UPFs, 70.6% vs. 66.7%, p = 0.473). However, most patients consumed diets high in ultra-processed foods. FI status was dynamic over the study period. Hospitalizations were more frequent among food-insecure patients. Conclusions: FI is common in pediatric IBD and associated with poorer nutritional status. FI was associated with higher consumption of UPFs, although diet quality was poor among most patients. Future studies should validate these findings in large cohorts and evaluate longitudinal interventions. Full article

Medical imaging analysis plays a pivotal role in modern healthcare, with physicians relying heavily on radiologists for disease diagnosis. However, many hospitals face a shortage of radiologists, leading to long queues at radiology centers and delays in diagnosis. Advances in artificial intelligence (AI) have made it possible for AI models to analyze medical images and provide insights similar to those of radiologists. Despite their successes, these models face significant challenges that hinder widespread adoption. One major issue is the inability of AI models to generalize data from new populations, as performance tends to degrade when evaluated on datasets with different or shifted distributions, a problem known as domain shift. Additionally, the large size of these models requires substantial computational resources for training and deployment. In this study, we address these challenges by investigating domain shifts using ChestXray-14 and a Nigerian chest X-ray dataset. We propose a multi-task learning (MTL) approach that jointly trains the model on both datasets for two tasks, classification and segmentation, to minimize the domain gap. Furthermore, we replace traditional convolutional layers in the backbone model (Densenet-201) architecture with depthwise separable convolutions, reducing the model’s number of parameters and computational requirements. Our proposed model demonstrated remarkable improvements in both accuracy and AUC, achieving 93% accuracy and 96% AUC when tested across both datasets, significantly outperforming traditional transfer learning methods. Full article

Early and accurate identification of brain tumors is essential for determining effective treatment strategies and improving patient outcomes. Artificial intelligence (AI) and deep learning (DL) techniques have shown promise in automating diagnostic tasks based on magnetic resonance imaging (MRI). This study evaluates the performance of four pre-trained deep convolutional neural network (CNN) architectures for the automatic multi-class classification of brain tumors into four categories: Glioma, Meningioma, Pituitary, and No Tumor. The proposed approach utilizes the publicly accessible Brain Tumor MRI Msoud dataset, consisting of 7023 images, with 5712 provided for training and 1311 for testing. To assess the impact of data availability, subsets containing 25%, 50%, 75%, and 100% of the training data were used. A stratified five-fold cross-validation technique was applied. The CNN architectures evaluated include DeiT3_base_patch16_224, Xception41, Inception_v4, and Swin_Tiny_Patch4_Window7_224, all fine-tuned using transfer learning. The training pipeline incorporated advanced preprocessing and image data augmentation techniques to enhance robustness and mitigate overfitting. Among the models tested, Swin_Tiny_Patch4_Window7_224 achieved the highest classification Accuracy of 99.24% on the test set using 75% of the training data. This model demonstrated superior generalization across all tumor classes and effectively addressed class imbalance issues. Furthermore, we deployed and benchmarked the best-performing DL model on embedded AI platforms (Jetson AGX Xavier and Orin Nano), demonstrating their capability for real-time inference and highlighting their feasibility for edge-based clinical deployment. The results highlight the strong potential of pre-trained deep CNN and transformer-based architectures in medical image analysis. The proposed approach provides a scalable and energy-efficient solution for automated brain tumor diagnosis, facilitating the integration of AI into clinical workflows. Full article

Gear faults are a frequent cause of rotating machinery breakdowns. There are two open issues in the current intelligent diagnosis model of gear faults. (1) Shallow models demand fewer data but necessitate feature extraction from raw signals, relying on prior knowledge. (2) Deep networks can adaptively extract fault features but require large datasets to train hyperparameters. In this paper, a novel fusion model, called CBAM-TCN-SVM, is proposed for intelligent gear fault diagnosis. It consists of a temporal convolutional network module (TCN), a convolutional block attention module (CBAM), and a support vector machine (SVM) module. More specifically, the frequency-domain sequence data are fed into the CBAM-TCN model, which effectively extracts deep fault features via multiple convolutional layers, channel attention mechanisms, and spatial attention mechanisms. Then, the SVM classifier is employed for intelligent classification. The fusion model combines the advantages of deep networks and shallow classifiers, addressing the issues that arise when the accuracy of fault diagnoses is constrained by the data scale and feature extractions rely on prior knowledge. The experiments result in the proposed method achieving a classification accuracy of 98.3% and demonstrate that it is a feasible approach for predicting gear faults. Full article

Persistent luminescence nanoparticles (PLNPs) represent a unique class of optical materials. They possess the ability to absorb and store energy from external excitation sources and emit light persistently once excitation terminates. Because of this distinctive property, PLNPs have attracted considerable attention in various areas. Especially in recent years, PLNPs have revealed marked benefits and extensive application potential in fields such as biological detection, imaging, targeted delivery, as well as integrated diagnosis and treatment. Not only do they potently attenuate autofluorescence interference arising from biological tissues, but they also demonstrate superior signal-to-noise ratio and sensitivity in in vivo imaging scenarios. Therefore, regarding the current research, this paper firstly introduces the classification, synthesis methods, and luminescence mechanism of the materials. Subsequently, the research progress of PLNPs in biological detection and imaging and medical treatment in recent years is reviewed. The challenges faced by materials in biomedical applications and the outlook of future development trends are further discussed, which delivers an innovative thought pattern for developing and designing new PLNPs to cater to more practical requirements. Full article

Background: Paroxysmal dyskinesias (PDs) are rare, episodic movement disorders characterized by sudden and involuntary hyperkinetic motor events. In paediatric populations, their diagnosis is often complicated by clinical overlap with epilepsy and other neurological conditions. Genetic underpinnings have increasingly been recognized as key to understanding phenotypic heterogeneity and guiding treatment. Objectives: This systematic review aims to provide a comprehensive overview of paediatric PD, with a focus on genetic aetiologies, clinical features, subtype classification, and therapeutic approaches, including genotype–treatment correlations. Methods: We systematically reviewed the literature from 2014 to 2025 using PubMed. Inclusion criteria targeted paediatric patients (aged 0–18 years) with documented paroxysmal hyperkinetic movements and genetically confirmed or clinically suggestive PD. Data were extracted regarding demographics, dyskinesia subtypes, age at onset, genetic findings, and treatment efficacy. Gene categories were classified as PD-specific or pleiotropic based on functional and clinical features. Results: We included 112 studies encompassing 605 paediatric patients. The most common subtype was Paroxistic Kinesigenic Dyskinesia (PKD). Male sex was more frequently reported. The mean onset age was 5.99 years. A genetic diagnosis was confirmed in 505 patients (83.5%), involving 38 different genes. Among these, PRRT2 was the most frequently implicated gene, followed by SLC2A1 and ADCY5. Chromosomal abnormalities affecting the 16p11.2 region were identified in ten patients, including deletions and duplications. Among the 504 patients with confirmed monogenic variants, 390 (77.4%) had mutations in PD-specific genes, while 122 (24.2%) carried pleiotropic variants. Antiseizure drugs—particularly sodium channel blockers such as carbamazepine and oxcarbazepine—were the most frequently reported treatment, with complete efficacy documented in 59.7% of the studies describing their use. Conclusions: Paediatric PDs exhibit significant clinical and genetic heterogeneity. While PRRT2 remains the most common genetic aetiology, emerging pleiotropic genes highlight the need for comprehensive diagnostic strategies. Sodium channel blockers are effective in a subset of genetically defined PD, particularly PRRT2-positive cases. Patients with pathogenic variants in other genes, such as ADCY5 and SLC2A1, may benefit from specific therapies that can potentially change their clinical course and prognosis. These findings support genotype-driven management approaches and underscore the importance of genetic testing in paediatric movement disorders. Full article

Background and Objectives: Spitz tumors represent a diagnostic challenge in dermatopathology due to their large spectrum of morphological characteristics and overlap with malignant lesions, especially in pathology departments where molecular pathology is not available. Even though most Spitz lesions are benign, the uncertainty around their biological behavior necessitates an integrated approach in daily practice. The objective of our study was to evaluate the epidemiological, macroscopic, and histopathological characteristics of Spitz lesions in accordance with WHO Classification of Skin Tumours. Materials and Methods: We performed a retrospective, descriptive, and hypothesis-generating study on Spitz tumors diagnosed between 2018 and 2024 in the Clinical Pathology Department of the Mures Clinical County Hospital, Romania. We included 10 cases and analyzed their macroscopic characteristics (localization, shape, dimension, and color), microscopic characteristics (cellular types, cytologic atypia, pagetoid migration, mitoses, and the type of lesion), and immunohistochemical profile. Results: The study population was composed of young patients with an average age of 20.2 years old, with a slight predominance of female gender. Most lesions were Spitz nevi, intradermic, or compound, with a fusiform, epithelioid, or rhomboid cell shape. Pagetoid migration and cytological atypia were seen in fewer cases. The Ki 67 proliferation index was under 5% in all cases. The main limitation of this study involved the low number of cases and the lack of molecular testing, which limited the molecular characterization of Spitz tumors. Complete excision was performed in all cases. Conclusions: In the absence of molecular testing, our study emphasizes the importance of clinical–morphological assessment using immunohistochemistry in establishing a correct diagnosis in Spitz lesions. Our results confirm that most of the Spitz lesions were benign and provide a basis for future research with a multidisciplinary approach, including molecular testing. Full article

Pancreatic Ductal Adenocarcinoma (PDAC) is one of the most lethal cancers, accounting for a significant proportion of cancer-related deaths globally. Despite advancements in medical science, treatment options for PDAC remain limited, and the prognosis is often poor. Early detection is a critical factor in improving patient outcomes, but current diagnostic methods often fail to detect PDAC until it has advanced to a late stage. In this context, the development of more effective diagnostic tools is of paramount importance. In this study, we explored the potential of non-coding RNAs (ncRNAs) as diagnostic markers for PDAC using cell-free nucleotides and liquid biopsies. Leveraging the power of Next Generation Sequencing (NGS), bioinformatics analysis, and machine learning (ML), we were able to identify unique RNA signatures associated with PDAC. Our findings revealed twenty key genes, including microRNAs (miRNAs), long-non-coding RNAs (lncRNAs), and miscellaneous RNAs that demonstrated high classification accuracy. Specifically, our model achieved a classification accuracy of 87% and an area under the receiver operating characteristic curve (AUC) of 91%. These ncRNAs could potentially serve as robust biomarkers for PDAC, offering a promising avenue for the development of a non-invasive diagnostic test. This could revolutionize PDAC diagnosis, enabling earlier detection and intervention, which is crucial for improving patient outcomes. This work lays the groundwork for future research, with the potential to significantly enhance PDAC diagnosis and therapy. Full article

Background: Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder characterized by social communication deficits and repetitive behaviors. Recent studies have suggested that gut microbiota may play a role in the pathophysiology of ASD. This study aims to develop a classification model for ASD diagnosis and to identify ASD-associated biomarkers by analyzing metagenomic data at the taxonomic level. Methods: The performances of five different methods were tested in this study. These methods are (i) SVM-RCE, (ii) RCE-IFE, (iii) microBiomeGSM, (iv) different feature selection methods, and (v) a union method. The last method is based on creating a union feature set consisting of the features with importance scores greater than 0.5, identified using the best-performing feature selection methods. Results: In our 10-fold Monte Carlo cross-validation experiments on ASD-associated metagenomic data, the most effective performance metric (an AUC of 0.99) was obtained using the union feature set (17 features) and the AdaBoost classifier. In other words, we achieve superior machine learning performance with a few features. Additionally, the SHAP method, which is an explainable artificial intelligence method, is applied to the union feature set, and Prevotella sp. 109 is identified as the most important microorganism for ASD development. Conclusions: These findings suggest that the proposed method may be a promising approach for uncovering microbial patterns associated with ASD and may inform future research in this area. This study should be regarded as exploratory, based on preliminary findings and hypothesis generation. Full article