Search Results (16,097)

The South African Government prioritises Small, Micro-, and Medium Enterprises (SMMEs) as catalysts for employment creation, in alignment with Sustainable Development Goal 8 (SDG 8), Decent Work and Economic Growth, which advocates for sustained, inclusive, and sustainable economic growth. However, the extent to which agro-processing SMMEs translate this policy ambition into measurable socio-economic gains remains contested due to persistent structural, financial, and operational constraints. This study develops a comprehensive, data-driven business support framework tailored to agro-processing SMMEs in the Free State province of South Africa. Employing a mixed-methods approach, survey data from 88 agro-processing SMMEs were analysed across 18 business performance dimensions. Average agreement scores and performance gaps were utilised to diagnose strengths and vulnerabilities within the sector. While overall performance was relatively strong (average agreement score: 86.7%), a critical weakness emerged in operational cost management (76.1%), revealing a 14.2% gap relative to the highest-performing dimension, equipment selection (90.3%). Based on these empirical insights, the study proposes a three-tiered business support architecture: (i) maintaining and leveraging high-performing dimensions (≥85% agreement), (ii) targeted enhancement for moderate-performing areas (80–84.9%), and (iii) crisis intervention for critical weaknesses (<80%). The framework integrates cross-cutting support services, including financing, regulatory guidance, and technology access, delivered through a phased implementation strategy comprising crisis intervention, system establishment, and optimisation and scaling. A multi-channel delivery mechanism, combining a hub-and-spoke model, mobile support units, and a digital platform, ensures provincial accessibility. By translating performance diagnostics into differentiated policy action, the framework promotes efficient resource allocation, supports both high-potential and vulnerable agro-processing SMMEs, and embeds a robust monitoring and evaluation system to track key performance indicators. The study contributes to the SMME development literature by demonstrating how structured, tiered, and context-specific support models can strengthen resilience, competitiveness, and sustainable agro-industrial growth in developing-country settings. Full article

Objectives: Ear involvement is a common feature of antineutrophil antibody (ANCA)-associated vasculitis (AAV). The vigilance of otolaryngologists often determines early diagnosis of AAV, thereby reducing the risk of irreversible organ damage and improving the quality of life of these patients. The goal of this study was to assess the quantitative and qualitative hearing impairment in patients with active AAV and to identify an audiological test that can detect of early deterioration of hearing even in patients without hearing loss. Methods: A study group of 46 patients with ANCA-associated vasculitis (AAV) hospitalized at the Department of Internal Diseases, Nephrology and Dialysis, Military Institute of Medicine—National Research Institute and a control group of 20 patients without a diagnosis of ANCA vasculitis and with no reported hearing disturbances were assessed prospectively. A battery of audiologic tests were carried out including High Frequency Audiometry, Impedance Audiometry, Otoacoustic Emissions, and Auditory Brainstem Responces (ABR). Computed Tomography of temporal bones and paranasal sinuses were performed, and audiologic anamnesis was gathered. Results: Pure-tone audiometry (PTA) demonstrated hearing loss in 58.6% (51/87) of the ears in the study group. The predominant type of damage was sensorineural hearing loss (SNHL). No correlation was found between hearing loss and AAV activity, duration of the disease, number of relapses, or ANCA antibody type. The statistically significant differences between control group and study group, even after excluding patients with hearing loss, were observed for high frequency audiometry (p < 0.001, for all tested frequencies excluding 14,000 Hz). The otoacoustic emissions showed to be statistically insignificant after exclusion of patients with hearing loss. Conclusions: Hearing involvement is common in patients with AAV regardless of the type of ANCA antibodies. High Frequency Audiometry could be an important audiologic screening test in this group of patients, and should be incorporated to diagnostic test battery in AAV. Otoacoustic emissions and ABR can be a handful in uncertain cases. Full article

Background/Objectives: Early differentiation of odontogenic cervicofacial necrotizing fasciitis (NF) from odontogenic abscess (OA) is clinically challenging yet critical due to the need for urgent surgical and antimicrobial escalation. We evaluated whether a novel C-reactive protein-to-platelet inflammatory index (CPII = CRP/platelets), combined with a symptom-based Symptom Severity (SS) score, improves early discrimination of NF from OA. Methods: This retrospective cohort study included 234 hospitalized patients with cervicofacial odontogenic infections treated between January 2010 and December 2023 (25 NF, 209 OA). Admission clinical variables, SS and SIRS scores, and laboratory parameters were analyzed. CPII and established immunoinflammatory indices (including AISI, SII, NLR, PLR, and LMR) were calculated. Group comparisons were performed using nonparametric and categorical tests. Diagnostic performance was assessed by ROC analysis, and multivariable logistic regression evaluated independent associations with NF. Results: Compared with OA, NF patients were older (median 42 [IQR 35–59] vs. 35 [IQR 26–49] years; p = 0.0098) and more frequently had comorbidities (52% vs. 25.4%; OR 3.19; p = 0.0087). Trismus and dysphagia were more common in NF (84% vs. 60.8%, p = 0.0272; 88% vs. 53.6%, p = 0.0010), with higher SS and SIRS scores (both p < 0.0001). NF was associated with longer hospitalization (median 17 vs. 6 days; p < 0.0001) and more complications (40% vs. 5.7%; OR 10.94; p < 0.0001). CRP was markedly higher in NF (median 287 vs. 111.5 mg/L; p < 0.0001), platelets were lower (median 210 vs. 249 × 10⁹/L; p = 0.0091), and CPII was substantially higher (median 1.23 vs. 0.45; p < 0.0001). AISI did not differ between groups (p = 0.861). ROC analysis demonstrated excellent discrimination for SS score (AUC 0.9328, cut-off 12), CRP (AUC 0.9109, cut-off 221 mg/L), and CPII (AUC 0.9271, cut-off 0.75), whereas AISI showed limited discrimination (AUC 0.5108). In multivariable analysis, both SS score (adjusted OR 2.08 per 1 point) and CPII (adjusted OR 6.87 per 0.5 units) were independently associated with NF; the combined SS + CPII model achieved an AUC of 0.9726. Conclusions: CPII is a simple, admission-available biomarker that differentiates odontogenic cervicofacial NF from OA with excellent accuracy and provides strong complementary value when combined with SS score. AISI, despite prior utility for odontogenic abscess severity assessment, did not discriminate NF from OA in this cohort. Full article

As critical components of wind energy systems, the structural integrity of wind turbine blades is directly tied to the operational safety and economic performance of wind turbines. With blade designs trending toward larger and more flexible structures and operating environments becoming increasingly harsh, maintenance strategies must urgently shift from reactive approaches to predictive maintenance paradigms. From an engineering application perspective, this study conducts a systematic and critical review of non-destructive testing (NDT) and structural health monitoring (SHM) technologies for wind turbine blades. Drawing on the literature published over the past decade, we examine the field applicability, limitations, and engineering challenges of core NDT techniques—including vision-based methods, acoustic approaches, vibration analysis, ultrasound, and infrared thermography. Particular emphasis is placed on the integration of data-driven approaches with engineering practice, evaluating the role of machine learning in fault classification and anomaly diagnosis, as well as the contributions of deep learning to automated defect detection in image and signal data. Moreover, this paper critically discusses the growing use of robotic inspection platforms, such as unmanned aerial vehicles and climbing robots, as multi-sensor carriers enabling rapid and comprehensive blade assessment. By comparatively analyzing detection performance, cost, and automation levels across technologies, we identify key engineering barriers, including environmental noise robustness, signal attenuation within complex blade structures, and the persistent gap between laboratory methods and field deployment. Finally, we outline forward-looking research directions, encompassing multi-modal sensor fusion, edge computing for real-time diagnostics, and the development of standardized SHM systems aimed at supporting full lifecycle blade management. Full article

Accurate detection of blood cells in microscopic images plays a crucial role in automated hematological analysis and clinical diagnosis. Herein, we proposed an improved YOLOv8n-based model for efficient and precise detection of red blood cells (RBCs), white blood cells (WBCs), and platelets in the BCCD dataset. The baseline YOLOv8n framework was enhanced by integrating GhostConv and C3Ghost modules to reduce model complexity while maintaining high detection performance. A series of ablation experiments were conducted to evaluate the individual and combined effects of these modules on model accuracy and computational efficiency. Experimental results demonstrated that the baseline model achieved an mAP@0.5 of 0.9043 with 3.01 M parameters. After incorporating GhostConv, the model maintained comparable accuracy (mAP@0.5 = 0.9040) with a reduction in parameters to 2.73 M. The C3Ghost integration further decreased parameters to 1.99 M with an mAP@0.5 of 0.8973. The combined model achieved an optimal balance between accuracy (mAP@0.5 = 0.9001) and compactness (1.71 M parameters). Results indicate that the improved YOLOv8n can effectively enhance detection efficiency without sacrificing precision. The proposed lightweight detection framework provides a promising solution for real-time blood cell analysis. Its high accuracy, reduced computational load, and strong generalization ability make it suitable for integration into automated laboratory systems, facilitating rapid and intelligent medical diagnostics in hematology and related biomedical applications. Full article

Background/Objectives: 18F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) is used in oncology but has limited specificity due to uptake in infectious and inflammatory conditions. This study evaluated the diagnostic value of 18F-FDG PET/CT in patients with infectious, inflammatory, and malignant lesions and its role in guiding histopathological biopsy decisions in an infectious disease setting. Methods: A retrospective cohort study included 186 adult patients who underwent 18F-FDG PET/CT between 2018 and 2023 for diagnostic evaluation at a tertiary care hospital. Clinical indications were fever of unknown origin (FUO), inflammation of unknown origin (IUO), lymphadenopathy of unknown origin, and suspected solid mass. Diagnostic yield, biopsy decisions, and factors associated with biopsy were analyzed. Results: The diagnostic yield of 18F-FDG PET/CT was 58.6%, with higher in malignant conditions (hematologic malignancy 85.7%, solid organ malignancy 88.9%) and lower in autoimmune/inflammatory diseases (20.8%) and mycobacterial infections. PET/CT showed moderate sensitivity (59.8%) and high specificity (98.7%) for infection detection, improving to 67.8% sensitivity after excluding mycobacterial infections. Biopsy was performed more in patients with lymphadenopathy, higher SUVmax (>7.4), and PET/CT findings not suggestive of infection. Analysis identified lymphadenopathy (aOR = 2.77), PET/CT not suggestive of infection (aOR = 4.73), and SUVmax > 7.4 (aOR = 4.98) as predictors of biopsy. Conclusions: 18F-FDG PET/CT provides moderate diagnostic value across infectious, inflammatory, and malignant diseases and guides biopsies effectively, particularly in patients with lymphadenopathy, elevated SUVmax, and non-infectious findings. Its limited performance in mycobacterial and autoimmune diseases requires cautious interpretation. Overall, 18F-FDG PET/CT supports clinical decisions in complex diagnostic scenarios. Full article

Robotic-assisted bronchoscopy (RAB) represents a significant technological advance, providing superior precision, enhanced visualization, and increased maneuverability relative to conventional bronchoscopic methods. This review provides an overview of current research evaluating RAB’s diagnostic performance and exploring future prospects. Recent literature demonstrates advantages in navigating difficult-to-reach lung lesions with improved safety profiles compared to transthoracic approaches. Incorporating advanced imaging technologies has enhanced real-time decision-making during procedures, and artificial intelligence applications are emerging. RAB has been rapidly adopted at many high-volume centers based on favorable navigational success and safety data. As the field matures, ongoing prospective studies will further define its role in improving patient outcomes, cost-effectiveness, and optimal integration with lung cancer screening programs. RAB faces ongoing challenges including substantial capital costs, training requirements, and need for standardized protocols. Therapeutic applications show promise and are under active investigation. Full article

Background/Objectives: Artificial intelligence (AI) tools, particularly large language models (LLMs) such as ChatGPT-4o, are gaining prominence in medicine. While their diagnostic capabilities have been explored across various oncologic domains, their role in clinical decision-making within multidisciplinary tumor boards (MTBs) remains largely unexamined in urologic oncology. This study evaluates the performance of ChatGPT-4o as a decision-support tool in a real-world MTB setting by comparing its recommendations with those of expert clinicians. Materials and Methods: A retrospective study was conducted using 98 anonymized clinical cases discussed by a urologic MTB between June 2024 and February 2025. An independent urologist entered the same cases into ChatGPT-4o using a standardized prompt replicating real-world presentation. Two certified urologists independently assessed the model’s responses. Agreement was analyzed overall and by tumor type, disease stage, clinical context, and treatment strategy. Results: ChatGPT-4o fully agreed with the MTB in 56.1% of cases, was correct but incomplete in 23.5%, and provided partially accurate but flawed recommendations in 18.4%. Overall concordance between ChatGPT-4o and the MTB yielded a Cohen’s kappa of 0.61, indicating moderate-to-good agreement. Discrepancies were most common in metastatic prostate cancer, often due to misclassification of tumor burden or errors in treatment sequencing. Highest agreement rates were observed in bladder and renal tumors, and in standardized therapeutic scenarios such as radiotherapy. Conclusions: ChatGPT-4o demonstrated moderate alignment with expert MTB decisions and performed best in well-defined clinical contexts. While it cannot replace multidisciplinary expertise, it may serve as a supportive tool to enhance access to standardized oncologic care. Full article

Electrical Impedance Tomography (EIT) represents a promising and non-invasive technique for the characterisation of biological tissues, but its diagnostic performance strongly depends on the electrode configuration, system geometry, and electronic acquisition strategies. In this work, a three-dimensional model based on the Finite Element Method (FEM) is developed to investigate the detectability of epithelial neoplasms through optimised electrode excitation schemes. The adjacent and opposite configurations are systematically compared in terms of impedance contrast, spatial sensitivity, and neoplastic inclusion localisation capability. The simulations were implemented using an open-source finite element solver with heterogeneous multilayer tissue models. The results show that the configuration with opposite electrodes significantly improves impedance contrast and sensitivity in three-dimensional models, allowing for better detection of localised conductivity anomalies. The proposed approach contributes to the design of optimised EIT electronic systems for early and non-invasive screening applications of epithelial cancer. Full article

Rumenitis is an inflammatory condition of the rumen, typically seen in adult cattle managed on high-energy diets. In calves, it is uncommon and often linked to ruminal drinking due to esophageal groove dysfunction. Early diagnosis is challenging due to nonspecific clinical signs. A one-month-old male Limousin calf was presented with persistent non-fetid fluid regurgitation, rhythmic mastication, inappetence, and progressive neurological signs. Clinical examination revealed signs of dehydration and neurological dysfunction. Laboratory evaluation demonstrated metabolic acidosis (pH 7.16), hyperkalemia, and elevated serum urea. Endoscopy identified diffuse mucosal hyperemia, erosions, and fluid accumulation in the rumen. Symptomatic and supportive therapy was initiated; however, the calf died spontaneously. Necropsy was therefore performed, and rumen samples were collected for histological and microbiological investigations. Histopathological analysis confirmed acute suppurative rumenitis. The microbiological culture of rumen and reticulum samples yielded mixed bacterial flora, including Escherichia coli and Proteus mirabilis. The fungal culture isolated Penicillium spp., Mucoraceae, Geotrichium spp., and Aspergillus fumigatus. This case details the value of integrating clinical examination, blood gas analysis, endoscopy, histopathology, and microbiology in diagnosing rumenitis in young calves. Although Limousin calves are not considered predisposed, management and feeding practices may play a critical role in disease onset. Rumenitis should be considered in calves presenting persistent regurgitation and neurological signs. Early, minimally invasive diagnostics such as endoscopy can improve diagnostic accuracy and inform timely clinical decision-making. Full article

Bone scintigraphy is a widely available and cost-effective modality for detecting skeletal metastases in prostate cancer, yet visual interpretation can be challenging due to heterogeneous uptake patterns, benign mimickers, and a high reporting workload, motivating robust computer-aided decision support. In this study, we present an experimental evaluation of fourteen convolutional neural network (CNN) architectures for binary metastasis classification in planar bone scintigraphy using a unified protocol. Fourteen models, CNN (baseline), AlexNet, VGG16, VGG19, ResNet18, ResNet34, ResNet50, ResNet50-attention, DenseNet121, DenseNet169, DenseNet121-attention, WideResNet50_2, EfficientNet-B0, and ConvNeXt-Tiny, were trained and tested on 600 scan images (300 normal, 300 metastatic) from the Jordanian Royal Medical Services under identical preprocessing and augmentation with stratified five-fold cross-validation. We report mean ± SD for AUC-ROC, accuracy, precision, sensitivity (recall), F1-score, specificity, and Cohen’s κ, alongside calibration via the Brier score and deployment indicators (parameters, FLOPs, model size, and inference time). DenseNet121 achieved the best overall balance of diagnostic performance and reliability, reaching AUC-ROC 96.0 ± 1.2, accuracy 89.2 ± 2.2, sensitivity 83.7 ± 3.4, specificity 94.7 ± 2.2, F1-score 88.5 ± 2.5, κ = 0.783 ± 0.045, and the strongest calibration (Brier 0.080 ± 0.013), with stable fold-to-fold behaviour. DenseNet121-attention produced the highest AUC-ROC (96.3 ± 1.1) but exhibited greater variability in specificity, indicating less consistent false-alarm control. Complexity analysis supported DenseNet121 as deployable (~7.0 M parameters, ~26.9 MB, ~92 ms/image), whereas heavier models yielded only limited additional clinical value. These results support DenseNet121 as a reliable backbone for automated metastasis detection in planar scintigraphy, with future work focusing on external validation, threshold optimisation, interpretability, and model compression for clinical adoption. Full article

Background/Objectives: Magnetite nanoparticles represent promising candidates for a broad spectrum of biomedical applications, ranging from in vitro diagnostic assays to in vivo imaging, hyperthermia, and targeted drug and gene delivery, with some nanoagents already approved for clinical use. A critical determinant of their functionality is the nanoparticle coating, which facilitates beneficial interactions within biological systems. In the context of tumor-targeted therapeutic delivery, key design parameters—particularly surface coatings—can be optimized to enhance treatment efficacy by modulating blood circulation kinetics, biodistribution, and other critical properties. However, current preclinical screening methods primarily rely on cell culture models to identify potential nanocarriers, yet these systems often poorly correlate with actual in vivo performance. This discrepancy highlights the necessity of incorporating more biologically relevant testing platforms, such as high-throughput in vivo assays. Methods: In this work, we employed an original magnetic particle quantification (MPQ) technology to systematically evaluate the blood circulation kinetics and biodistribution patterns for magnetite nanoparticles with 17 different coatings across multiple organs and tissues, including the liver, spleen, lungs, kidneys, heart, tumor, brain, peripheral blood, muscle, and bone. This methodology offers high sensitivity, user-friendly operation, and provides quantitative measurements across a broad dynamic range of nanoparticle concentrations. These advantages enabled high-throughput acquisition of precise blood circulation and biodistribution data. In addition, histological analysis was conducted to evaluate nanoparticle penetration depth within tumor tissue. Results: Here we conducted a comprehensive study of the effect of 17 different polymer-, lectin-, and small molecule-based coatings on the behavior of magnetite nanoparticles in vivo. For each type of obtained nanoparticles, we implemented passive targeting as well as magnetic targeting, the latter using an external magnetic field localized in the tumor area. Conclusions: The collected dataset provides critical insights into how surface modifications influence nanoparticle performance in complex biological systems, offering valuable guidance for optimizing therapeutic nanocarrier design. Full article

Background/Objectives: Although Ki-67 is not formally incorporated into the grading system of pulmonary neuroendocrine neoplasms (PNENs), it is widely used as an adjunct marker to reflect proliferative activity and support diagnostic stratification. Manual Ki-67 assessment is subject to interobserver variability and methodological limitations. This study aimed to evaluate the reliability and performance of two artificial intelligence (AI)-based image analysis systems in Ki-67 index assessment and to compare their results with expert pathologist evaluation in pulmonary neuroendocrine tumors. Methods: A total of 63 pulmonary neuroendocrine neoplasm cases, including typical carcinoid (n = 29), atypical carcinoid (n = 13), and large cell neuroendocrine carcinoma (n = 21), were retrospectively analyzed. Ki-67 proliferation indices were independently assessed by four pathologists within predefined hotspot regions, counting approximately 2000 tumor cells per case. The same regions were analyzed using two AI-based image analysis systems (Roche uPath Ki-67 and Virasoft Virasight Ki-67). Interobserver agreement among pathologists was evaluated using the intraclass correlation coefficient (ICC), and concordance between manual and AI-based assessments was assessed using Spearman’s correlation and linear regression analyses. To account for potential scanner/platform effects, slides were digitized using two different whole-slide scanners (VENTANA DP^® 600 and Leica Aperio AT2), and color normalization and quality control procedures were applied prior to AI-based analysis. For clinical interpretability, Ki-67 indices were stratified into categorical groups based on tumor subtype-specific thresholds (0–<10%: low, 10–25%: intermediate, >25%: high), and agreement between manual and AI-based categorical scoring was evaluated using Cohen’s kappa coefficient. Results: Among the 63 pulmonary neuroendocrine neoplasm cases, Ki-67 proliferation indices varied across tumor subtypes, with typical carcinoids showing low, atypical carcinoids intermediate, and large cell neuroendocrine carcinomas high proliferative activity. Interobserver agreement among four pathologists was excellent (ICC = 0.998, 95% CI: 0.996–0.998). Strong correlations were observed between manual Ki-67 assessments and AI-derived indices, with Spearman correlation coefficients of 0.961 (95% CI: 0.918–0.982) for Roche AI and 0.904 (95% CI: 0.821–0.949) for Virasoft AI, and 0.926 (95% CI: 0.842–0.968) between the two AI systems. Bland–Altman analyses demonstrated minimal mean differences and most cases within the 95% limits of agreement, indicating high concordance without systematic bias. Categorical agreement analysis, using subtype-specific Ki-67 thresholds (0–<10%: low; 10–25%: intermediate; >25%: high), showed excellent concordance between manual and AI-based scoring (Cohen’s kappa 0.877 for Roche AI and 0.827 for Virasoft AI; p < 0.001), confirming the clinical interpretability and reproducibility of AI-based Ki-67 assessment. Conclusions: AI-based Ki-67 index assessment shows strong concordance with expert pathologist evaluation and reflects biologically relevant differences among pulmonary neuroendocrine neoplasm subtypes. These results suggest that AI-assisted Ki-67 analysis may serve as a reproducible and objective adjunct to routine diagnostic practice in pulmonary neuroendocrine tumors. Full article

Sexually transmitted infections (STIs) remain a major global public health concern, with more than one million new cases acquired daily and increasing antimicrobial resistance compromising effective control strategies. Rapid, accurate and multiplex molecular diagnostics are therefore essential to support timely clinical management and public health surveillance. This study evaluated the analytical and clinical performance of the cartridge-based STANDARD™ M10 STI Panel (SD Biosensor, Republic of Korea), a fully automated, random-access real-time PCR assay capable of detecting eight STI-related pathogens within approximately 64 min. A total of 150 residual clinical specimens were retrospectively analysed, including vaginal, rectal, urethral and oropharyngeal swabs, as well as seminal fluids, and results were compared with those obtained using the Allplex™ STI Essential Assay (Seegene, Republic of Korea), which shares six common targets. The STANDARD™ M10 STI Panel demonstrated high diagnostic accuracy, with sensitivity ranging from 89.5% to 100%, specificity from 98.2% to 100% and overall accuracy between 98% and 100%. Agreement between the two assays was almost perfect, with Cohen’s κ values ranging from 0.91 to 1.00. Analytical sensitivity was further confirmed through verification of the limits of detection using quantified reference standards. Although validated for urine samples, the assay also showed robust performance on alternative clinical matrices, particularly vaginal swabs. Overall, these findings indicate that the STANDARD™ M10 STI Panel represents a reliable and practical tool for STI diagnosis, combining rapid turnaround time, minimal hands-on requirements and broad pathogen coverage in both centralized and near-patient testing settings. Full article

Artificial intelligence technology offers an intelligent and efficient new pathway for bearing fault diagnosis, holding significant importance for ensuring the stable operation of industrial systems. However, bearing fault samples are scarce in industrial practice, and traditional data-driven methods exhibit a marked decline in diagnostic performance under conditions of small sample sizes. To address this, this paper proposes a few-shot bearing fault diagnosis method that integrates an Improved Generative Adversarial Network with a CNN-BiLSTM-Attention hybrid network. The method comprises three core stages: in the data augmentation stage, a class-center-constrained Least Squares Generative Adversarial Network (CCC-LSGAN) model featuring class center constraint and joint loss optimization is proposed to generate high-quality fault samples through frequency-domain feature constraints, effectively expanding the training data; in the feature learning stage, a one-dimensional Convolutional Neural Network, Bidirectional Long Short-Term Memory, and Attention hybrid network (1D-CNN-BiLSTM-Attention) hybrid base classifier is constructed, which combines multi-scale convolution, bidirectional temporal modeling, and attention mechanisms to fully extract the spatiotemporal features of vibration signals; in the inference stage, test-time noise augmentation and a multi-model weighted voting ensemble mechanism are introduced to enhance the robustness and generalization capability of the diagnosis. Experimental results based on the PU and CWRU public bearing datasets demonstrate that the proposed method significantly outperforms existing mainstream diagnostic approaches in core metrics, including accuracy, precision, recall, and F1 score. It achieves a diagnostic accuracy of 96.60% on the PU dataset and 98.58% on the CWRU dataset. This method verifies the feasibility of highly reliable diagnosis under few-shot conditions and provides an effective solution for the intelligent operation and maintenance of industrial equipment. Full article