Browse Articles
Discover research articles across all indexed journals
Comparison between norepinephrine plus epinephrine and norepinephrine plus vasopressin after return of spontaneous circulation in patients with out-of-hospital cardiac arrest
Identification of unique biomarkers for proliferative diabetic retinopathy with tractional retinal detachment by proteomics profiling of vitreous humor
AI-driven cybersecurity framework for software development based on the ANN-ISM paradigm
Abstract With the increasing reliance on software applications, cybersecurity threats have become a critical concern for developers and organizations. The answer to this vulnerability is AI systems, which help us adapt a little better, as traditional measures in security have failed to respond to the upcoming threats. This paper presents an innovative cybersecurity framework using AI, by the Artificial Neural Network (ANN)—Interpretive Structural Modeling (ISM) model, to improve threat detection, vulnerability assessment, and risk response during software development. This framework helps realize dynamic, intelligent security as a part of the Software Development life cycle (SDLC). Initially, existing cybersecurity risks in software coding are systematically evaluated to identify potential gaps and integrate best practices into the proposed model. In the second phase, an empirical survey was conducted to identify and validate the findings of the systematic literature review (SLR). In the third phase, a hybrid approach is employed, integrating ANN for real-time threat detection and risk assessment. It utilizes ISM to analyze the relationships between cybersecurity risks and vulnerabilities, creating a structured framework for understanding interdependencies. A case study was conducted in the last stage to test and evaluate the AI-driven cybersecurity Mitigation Model for Secure Software Coding. A multi-level categorization system is also used to assess maturity across five key levels: Ad hoc, Planned, Standardized, Metrics-Driven, and Continuous Improvements. This study identifies 15 cybersecurity risks and vulnerabilities in software coding, along with 158 AI-driven best practices for mitigating these risks. It also identifies critical areas of insecure coding practices and develops a scalable model to address cybersecurity risks across different maturity levels. The results show that AI outperforms traditional systems in detecting security weaknesses and simultaneously fixing problems. During Levels 1–3 of the system improvement process, advanced security methods are used to protect against threats. Our analysis reveals that organizations at Levels 4 and 5 still need to entirely shift to using AI-based protection tools and techniques. The proposed system provides developers and managers with valuable insights, enabling them to select security enhancements tailored to their organization's development stages. It supports automated threat analysis, helping organizations stay vigilant against potential cybersecurity threats. The study introduces a novel ANN-ISM framework integrating AI tools with cybersecurity modeling formalisms. By merging AI systems with secure software coding principles, this research enhances the connection between AI-generated insights and real-world cybersecurity usage.
TSTA-GCN: trend spatio-temporal traffic flow prediction using adaptive graph convolution network
Ensemble stacked model for enhanced identification of sentiments from IMDB reviews
A prospective cohort study on the joint associations of abdominal aortic calcification and systemic inflammation response index with mortality risk
Influence of air pollution and climate variability on dengue in Singapore: a time-series analysis
Mortality among individuals with chronic kidney disease based on the 2012 and 2021 KDIGO blood pressure targets
Validity of tremor analysis using smartphone compatible computer vision frameworks
Abstract Computer vision (CV)-based approaches hold promising potential for the classification and quantitative assessment of movement disorders. To take full advantage of this potential, the pipelines need to be validated against established clinical and electrophysiological gold standards. This study examines the validity of the Mediapipe (by Google) and Vision (by Apple) smartphone-enabled hand detection frameworks for tremor analysis. Both frameworks were tested in virtual experiments with simulated tremulous hands to determine the optimal camera position for hand tremor assessment and the minimum detectable tremor amplitude and frequency. Both frameworks were then compared with optical motion capture (OMC), accelerometry, and clinical ratings in 20 tremor patients. Both CV frameworks accurately measured tremor peak frequency. Significant correlations were found between CV-assessed tremor amplitudes and Essential Tremor Rating Assessment Scale (TETRAS) scores. However, the accuracy of amplitude estimation compared to OMC as ground truth was insufficient for clinical application. In conclusion, CV-based tremor analysis is an accurate and simple clinical assessment tool to determine tremor frequency. Further improvements in amplitude estimation are needed.
Characterization of zebrafish rod and cone photoresponses
Abstract Zebrafish is a popular species widely used in vision research. The zebrafish retina has one rod and four cone subtypes (UV-, blue-, green-, and red-sensitive) with 40%-rod 60%-cone ratio, making it suitable for comparable studies of rods and cones in health and disease. However, the basic photoresponse properties of the four zebrafish cone subtypes have not been described yet. Here, we established a method for collecting flash photoresponses from zebrafish rods and cones by recording membrane current with a suction electrode. Photoreceptor subtypes could be distinguished based on their characteristic morphology and spectral sensitivity. Rods showed 40–220-fold higher photosensitivity than cones. In the four cone subtypes, green-sensitive cones showed the highest sensitivity, 5.5-fold higher than that of red cones. Unexpectedly, rods produced smaller flash responses than cones despite their larger outer segments. Dim flash response analysis showed the quickest response kinetics in blue- and red-sensitive cones, with responses about 2-fold faster than the responses of UV- and green-sensitive cones, and 6.6-fold faster than the rod responses. We also obtained pharmacologically isolated photoreceptor voltage responses (a-wave) from isolated zebrafish retinas using ex vivo electroretinography (ERG). Dim flashes evoked rod-only responses, while bright flashes evoked two-component responses with a slow rod component and a fast cone component. Red- and green-sensitive cones were the dominant sources of the overall cone response. These studies provide a foundation for the use of zebrafish rods and cones to study the fundamental mechanisms that modulate the function of vertebrate photoreceptors in health and disease.
A prediction method for radiation proctitis based on SAM-Med2D model
Abstract Cervical cancer, a prevalent gynecological malignancy, poses significant threats to women’s health. Despite advances in treatment modalities, radiotherapy remains a cornerstone in managing cervical cancer. However, radiotherapy-induced complications, such as radiation proctitis, present substantial diagnostic and prognostic challenges. Accurate diagnosis are crucial for optimizing treatment strategies and improving patient outcomes. Deep learning has shown remarkable success in medical image segmentation, aiding clinicians in assessing patient conditions. In the other hand, radiomics excels in extracting diagnostically valuable features from medical images but requires extensive manual annotation and often lacks generalizability. Therefore, combining the strengths of deep learning and radiomics is pivotal in addressing these challenges. In this study, we propose a novel paradigm that leverages deep learning models for initial segmentation, followed by detailed radiomics analysis. Specifically, we utilize the Transformer-based SAM-Med2D model to extract visual features from CT images of cervical cancer patients. We apply T-tests and Lasso regression to identify features most correlated with radiation proctitis and build predictive models using logistic regression, random forest, and naive Gaussian Bayesian algorithms. Experimental results demonstrate that our method effectively extracts CT imaging features and exhibits excellent performance in diagnosis radiation proctitis. This approach not only enhances predictive accuracy but also provides a valuable tool for personalizing treatment plans and improving patient outcomes in cervical cancer radiotherapy.
Two antagonistic objectives for one multi-scale graph clustering framework
The influence of section diameter on the ultrasonic fatigue response of 316L stainless steel manufactured via laser powder bed fusion
Abstract In this investigation, the influence of section diameter on high cycle fatigue (HCF) behavior of additively manufactured 316 L stainless steel was characterized. Three gauge-section diameters (5.0 mm, 2.5 mm, and 1.5 mm) were examined for their influence on the ultrasonic fatigue response of samples built via laser-powder bed fusion (L-PBF). HCF was conducted under full reversed loading ( R=−1 ) conditions. A total of 130 specimens were characterized in the as-built state at maximum stresses ranging from 70 to 220 MPa. A Random Fatigue Limit (RFL) model using a Maximum Likelihood Estimation (MLE) was used to quantify statistical variability and estimate an S-N curve fit. The fatigue response shows that the largest gauge diameter (5.0 mm) resulted in the lowest fatigue strength at 89.5 ± 5.6 MPa, and the smallest diameter (1.5 mm) resulted in the highest fatigue strength at 122.0 ± 32.8 MPa. The 2.5 mm diameter specimens exhibited a fatigue strength of 98.7 ± 7.0 MPa. The primary failure mechanism in all as-built specimens was surface initiated cracking from crevices in the as-built surface finish. Additional specimens with a nominal diameter of 5.0 mm were fatigue tested with the as-built surface removed via low stress surface grinding. The fatigue strength of these samples increased to 170 MPa when 75 μm of the surface was removed and 179 MPa when the surface contour was entirely removed. Residual stresses were characterized by x-ray diffraction (XRD) and show a reduced axial residual stress with reduction in gauge diameter. Additional specimens were fatigue tested after undergoing a stress relief anneal, resulting in a 51% reduction in the residual stress and a 30% improvement in fatigue strength. An in-depth analysis of the microstructure, surface roughness, defects, and fracture surface indicate that both the surface condition and residual stress are the primary factors influencing the observed diameter effects on HCF.
Weight-adjusted-waist index is associated with urinary albumin-creatinine ratio in normal body mass index adults: a cross-sectional study from NHANES 2001–2018
Enhancing faculty teaching performance through constructive leadership with a mediating role of job satisfaction
Association between central obesity and ADL impairment among the middle-aged and elderly population in China based on CHARLS
Extraction of kaolin and tribo informative analysis of the Al-kaolin composite through machine learning approaches
Exploring a multi-path U-net with probability distribution attention and cascade dilated convolution for precise retinal vessel segmentation in fundus images
Abstract While deep learning has become the go-to method for image denoising due to its impressive noise removal Retinal blood vessel segmentation presents several challenges, including limited labeled image data, complex multi-scale vessel structures, and susceptibility to interference from lesion areas. To confront these challenges, this work offers a novel technique that integrates attention mechanisms and a cascaded dilated convolution module (CDCM) within a multi-path U-Net architecture. First, a dual-path U-Net is developed to extract both coarse and fine-grained vessel structures through separate texture and structural branches. A CDCM is integrated to gather multi-scale vessel features, enhancing the model’s ability to extract deep semantic features. Second, a boosting algorithm that incorporates probability distribution attention (PDA) within the upscaling blocks is employed. This approach adjusts the probability distribution, increasing the contribution of shallow information, thereby enhancing segmentation performance in complex backgrounds and reducing the risk of overfitting. Finally, the output from the dual-path U-Net is processed through a feature refinement module. This step further refines the vessel segmentation by integrating and extracting relevant features. Results from experiments on three benchmark datasets, including CHASEDB1, DRIVE, and STARE, demonstrate that the proposed method delivers improved segmentation accuracy compared to existing techniques.