Structure Of Dataset Research Articles

A TMEM260 biallelic deletion underlies truncus arteriosus

Introduction: Truncus arteriosus (TA) is a life-threatening cardiovascular anomaly involving a ventricular septal defect and a common ventricular outflow tract. Recently, biallelic variants in TMEM260 have been identified as causative for Structural Heart Defects and Renal Anomalies syndrome (SHDRA, MIM 617478), which includes TA. Approximately 30 patients with SHDRA have been reported, but the genotype–phenotype correlation remains unclear. Founder variants have been identified in patients of East Asian and Ashkenazi Jewish ancestry. Case Presentation: The male infant, the third child of unrelated Japanese parents, was prenatally diagnosed with TA via detailed ultrasound examination. His older sister also had TA and died at 20 days of age. Despite intensive cardiorespiratory care, the patient passed away at 53 days due to heart failure. Genetic analysis identified a homozygous deletion of TMEM260 exons 6 and 7, resulting from a 7066 bp deletion with a 1 bp insertion, inherited from heterozygous parents. This deletion is included in the structural variation dataset with an allele frequency of 0.00173 (29/8380) in ToMMo 8.3K JPN-SV. Discussion: This is the first familial TA caused by a biallelic structural variation in TMEM260. The 7066-bp deletion shows a high allele frequency in the Japanese population. These findings support the idea that this TMEM260 deletion may be a significant cause of TA and should therefore be considered alongside the previously known c.1617del variant as a potential causative factor.

An in silico framework to visualize how cancer-associated mutations influence structural plasticity of the chemokine receptor CCR3.

G protein Coupled Receptors (GPCRs) are the largest family of cell surface receptors in humans. Somatic mutations in GPCRs are implicated in cancer progression and metastasis, but mechanisms are poorly understood. Emerging evidence implicates perturbation of intra-receptor activation pathway motifs whereby extracellular signals are transmitted intracellularly. Recently, sufficiently sensitive methodology was described to calculate structural strain as a function of missense mutations in AlphaFold-predicted model structures, which was extensively validated on experimental and predicted structural datasets. When paired with Molecular Dynamics (MD) simulations, these tools provide a facile approach to screen mutations in silico. We applied this framework to calculate the structural and dynamic effects of cancer-associated mutations in the chemokine receptor CCR3, a Class A GPCR involved in cancer and autoimmune disorders. Residue-residue contact scoring refined effective strain results, highlighting significant remodeling of inter- and intra-motif contacts along the highly conserved GPCR activation pathway network. We then integrated AlphaFold-derived predicted Local Distance Difference Test scores with per-residue Root Mean Square Fluctuations and activation pathway Contact Analysis (CONAN) from coarse grain MD simulations to identify statistically significant changes in receptor dynamics upon mutation. Finally, analysis of negative control mutants suggests false positive results in AlphaFold pipelines should be considered but can be mitigated with stricter control of statistical analysis. Our results indicate selected mutants influence structural plasticity of CCR3 related to ligand interaction, activation, and G protein coupling, using a framework that could be applicable to a wide range of biochemically relevant protein targets following further validation.

Enhance the Concrete Crack Classification Based on a Novel Multi-Stage YOLOV10-ViT Framework.

Early identification of concrete cracks and multi-class detection can help to avoid future deformation or collapse in concrete structures. Available traditional detection and methodologies require enormous effort and time. To overcome such difficulties, current vision-based deep learning models can effectively detect and classify various concrete cracks. This study introduces a novel multi-stage deep learning framework for crack detection and type classification. First, the recently developed YOLOV10 model is trained to detect possible defective regions in concrete images. After that, a modified vision transformer (ViT) model is trained to classify concrete images into three main types: normal, simple cracks, and multi-branched cracks. The evaluation process includes feeding concrete test images into the trained YOLOV10 model, identifying the possible defect regions, and finally delivering the detected regions into the trained ViT model, which decides the appropriate crack type of those detected regions. Experiments are conducted using the individual ViT model and the proposed multi-stage framework. To improve the generation ability, multi-source datasets of concrete structures are used. For the classification part, a concrete crack dataset consisting of 12,000 images of three classes is utilized, while for the detection part, a dataset composed of various materials from historical buildings containing 1116 concrete images with their corresponding bounding boxes, is utilized. Results prove that the proposed multi-stage model accurately classifies crack types with 90.67% precision, 90.03% recall, and 90.34% F1-score. The results also show that the proposed model outperforms the individual classification model by 10.9%, 19.99%, and 19.2% for precision, recall, and F1-score, respectively. The proposed multi-stage YOLOV10-ViT model can be integrated into the construction systems which are based on crack materials to obtain early warning of possible future deformation in concrete structures.

A dataset of Zostera marina and Zostera noltei structure and functioning in four sites along the French coast over a period of 18 months

A dataset of Zostera marina and Zostera noltei structure and functioning in four sites along the French coast over a period of 18 months

Data science shows that entropy correlates with accelerated zeolite crystallization in Monte Carlo simulations.

We have performed a data science study of Monte Carlo (MC) simulation trajectories to understand factors that can accelerate the formation of zeolite nanoporous crystals, a process that can take days or even weeks. In previous work, MC simulations predicted and experiments confirmed that using a secondary organic structure-directing agent (OSDA) accelerates the crystallization of all-silica LTA zeolite, with experiments finding a three-fold speedup [Bores et al., Phys. Chem. Chem. Phys. 24, 142-148 (2022)]. However, it remains unclear what physical factors cause the speed-up. Here, we apply data science to analyze the simulation trajectories to discover what drives accelerated zeolite crystallization in MC simulations going from a one-OSDA synthesis (1OSDA) to a two-OSDA version (2OSDA). We encoded simulation snapshots using the smooth overlap of atomic positions approach, which represents all two- and three-body correlations within a given cutoff distance. Principal component analyses failed to discriminate datasets of structures from 1OSDA and 2OSDA simulations, while the Support Vector Machine (SVM) approach succeeded at classifying such structures with an area-under-curve (AUC) score of 0.99 (where AUC = 1 is a perfect classification) with all three-body correlations and as high as 0.94 with only two-body correlations. SVM decision functions reveal relatively broad/narrow histograms for 1OSDA/2OSDA datasets, suggesting that the two simulations differ strongly in information heterogeneity. Informed by these results, we performed pair (2-body) entropy calculations during crystallization, resulting in entropy differences that semi-quantitatively account for the speedup observed in the previous MC simulations. We conclude that altering synthesis conditions in ways that substantially change the entropy of labile silica networks may accelerate zeolite crystallization, and we discuss possible approaches for achieving such acceleration.

Deep knowledge transfer powered ultrasonic guided wave damage monitoring under incomplete database scenarios: Theories, applications and challenges

Abstract Ultrasonic guided waves can travel long distances within the detected structures, which is of great significance for monitoring large complex engineering systems. However, the multimodal and dispersive properties of the specific research object making this promising whole structure monitoring difficult to interpret the signal mathematically and physically. With the development and maturity of deep learning and big data mining technologies, many scholars have noticed artificial intelligence algorithms such as deep learning can provide a new tool in ultrasonic guided wave signal processing, avoiding the mechanism analysis difficulties in the application of ultrasonic guided wave. But the integrity of structural state data sets has become a new pain point in engineering applications under this new approach, and how to apply the knowledge obtained from the existing data set to different but related fields through knowledge transfer in such cases begin to attract the attention of scholars and engineers. Although several systematic and valuable review articles on data-driven ultrasonic guided wave monitoring methods have been published, they only summarized relevant studies from the perspective of data-driven algorithms, ignoring the knowledge transfer process in practical application scenarios, and the intelligent ultrasonic guided wave monitoring methods based on knowledge transfer of incomplete sets are still lacking a comprehensive review. This paper focuses on the ultrasonic guided wave transfer monitoring technology when the training sample is missing, explores the feature correlation between samples in different domains, improves the transfer ability of the structural monitoring model under different conditions, and analyzes the ultrasonic guided wave intelligent monitoring methods for structural state under different sample missing conditions from three aspects: semi-supervised monitoring, multi-task transfer and cross-structure transfer. It is also expected to provide a new method and approach to solve the condition monitoring problems in other complex scenarios.

A Study of New Approaches to Statistical Analysis of Research data

Statistical analysis has been at the heart of scientific research, providing critical tools for data interpretation, decision-making, and hypothesis testing. Some of the ancient techniques used were hypothesis testing, regression analysis, and time series analysis, among others. Such methods have proved to be good tools for researchers dealing with smaller and more structured datasets. However, the large size and complexity of the dataset exposed the weaknesses in such classical approaches, especially in handling large, unstructured, or non-linear datasets. Increased computing power and the development of machine learning algorithms have increased flexibility, nudging the statistical approach towards more flexible, data-driven methods. This paper reviews emerging approaches such as machine learning, deep learning, Bayesian methods, and network analysis and places an emphasis on how these approaches can be applied over a range of fields where they could vastly transform statistical analysis. Based on this review, comparing traditional and modern methodologies, it thereby demonstrates how innovations better complement rather than replacing the capabilities of statistical analysis, thus shaping the future of research in this changing environment.

Protein engineering using variational free energy approximation

Engineering proteins is a challenging task requiring the exploration of a vast design space. Traditionally, this is achieved using Directed Evolution (DE), which is a laborious process. Generative deep learning, instead, can learn biological features of functional proteins from sequence and structural datasets and return novel variants. However, most models do not generate thermodynamically stable proteins, thus leading to many non-functional variants. Here we propose a model called PRotein Engineering by Variational frEe eNergy approximaTion (PREVENT), which generates stable and functional variants by learning the sequence and thermodynamic landscape of a protein. We evaluate PREVENT by designing 40 variants of the conditionally essential E. coli phosphotransferase N-acetyl-L-glutamate kinase (EcNAGK). We find 85% of the variants to be functional, with 55% of them showing similar growth rate compared to the wildtype enzyme, despite harbouring up to 9 mutations. Our results support a new approach that can significantly accelerate protein engineering.

The development of classification-based machine-learning models for the toxicity assessment of chemicals associated with plastic packaging

Assessing chemical toxicity in materials like plastic packaging is critical to safeguarding public health. This study presents the development of classification-based machine learning models to predict the toxicity of chemicals associated with plastic packaging. Using an extensive dataset of chemical structures, we trained multiple machine learning models—Random Forest, Support Vector Machine, Linear Discriminant Analysis, and Logistic Regression—targeting endpoints such as Neurotoxicity, Hepatotoxicity, Dermatotoxicity, Carcinogenicity, Reproductive Toxicity, Skin Sensitization, and Toxic Pneumonitis. The dataset was pre-processed by selecting 2D molecular descriptors as feature inputs, with resampling methods (ADASYN, Borderline SMOTE, Random Over-sampler, SVMSMOTE Cluster Centroid, Near Miss, Random Under Sampler) applied to balance classes for accurate classification. A five-fold cross-validation technique was used to optimize model performance, with model parameters fine-tuned using grid search. The model performance was evaluated using accuracy (Acc), sensitivity (Se), specificity (Sp), and area under the receiver operating characteristic curve (AUC-ROC) metrics. In most of the cases, the model accuracy was 0.8 or above for both training and test sets. Additionally, SHAP (SHapley Additive exPlanations) values were utilized for feature importance analysis, highlighting significant descriptors contributing to toxicity predictions. The models were ranked using the Sum of Ranking Differences (SRD) method to systematically select the most effective model. The optimal models demonstrated high predictive accuracy and interpretability, providing a scalable and efficient solution for toxicity assessment compared to traditional methods. This approach offers a valuable tool for rapidly screening potentially hazardous chemicals in plastic packaging.

Structural Damage Early Warning Method of Quayside Container Crane Based on Fuzzy Entropy Ratio Variation Deviation.

Real-time monitoring and early warning of structures are essential for assessing structural health and ensuring safety maintenance. To improve the timeliness of early warnings for structural abnormal states in quayside container cranes (QCCs) with incomplete damage data, a structural abnormal state early warning method based on fuzzy entropy ratio variation deviation (FERVD) is proposed. First, monitoring data are subjected to dual-tree complex wavelet transform (DTCWT). The adaptive frequency bands obtained from the decomposition, combined with fuzzy entropy (FE), are used to extract response signal features and construct the FERVD warning indicator. Based on this indicator, dynamic thresholds for early warning are established to differentiate between structural health states and various damage conditions. Secondly, a finite element model of structure for QCCs is developed. By simulating damage at various locations and severities through the stiffness reduction of different elements, a comprehensive structural simulation monitoring dataset is generated. The efficacy of the proposed early warning method is validated through numerical experiments and engineering case studies. The numerical results demonstrate that the proposed method effectively distinguishes between different damage conditions and provides timely warnings for various damage states. Furthermore, engineering case analysis shows that when the structure is in a healthy state, the FERVD values at different monitoring points fluctuate within the threshold range, indicating the applicability of the proposed method in the structural health monitoring (SHM) of QCCs.

FL-YOLOv8: Lightweight Object Detector Based on Feature Fusion

In recent years, anchor-free object detectors have become predominant in deep learning, the YOLOv8 model as a real-time object detector based on anchor-free frames is universal and influential, it efficiently detects objects across multiple scales. However, the generalization performance of the model is lacking, and the feature fusion within the neck module overly relies on its structural design and dataset size, and it is particularly difficult to localize and detect small objects. To address these issues, we propose the FL-YOLOv8 object detector, which is improved based on YOLOv8s. Firstly, we introduce the FSDI module in the neck, enhancing semantic information across all layers and incorporating rich detailed features through straightforward layer-hopping connections. This module integrates both high-level and low-level information to enhance the accuracy and efficiency of image detection. Meanwhile, the structure of the model was optimized and designed, and the LSCD module is constructed in the detection head; adopting a lightweight shared convolutional detection head reduces the number of parameters and computation of the model by 19% and 10%, respectively. Our model achieves a comprehensive performance of 45.5% on the COCO generalized dataset, surpassing the benchmark by 0.8 percentage points. To further validate the effectiveness of the method, experiments were also performed on specific domain urine sediment data (FCUS22), and the results on category detection also better justify the FL-YOLOv8 object detection algorithm.

Anonymize or synthesize? Privacy-preserving methods for heart failure score analytics

Abstract Aims Data availability remains a critical challenge in modern, data-driven medical research. Due to the sensitive nature of patient health records, they are rightfully subject to stringent privacy protection measures. One way to overcome these restrictions is to preserve patient privacy by using anonymization and synthetization strategies. In this work, we investigate the effectiveness of these methods for protecting patient privacy using real-world cardiology health records. Methods and results We implemented anonymization and synthetization techniques for a structure data set, which was collected during the HiGHmed Use Case Cardiology study. We employed the data anonymization tool ARX and the data synthetization framework ASyH individually and in combination. We evaluated the utility and shortcomings of the different approaches by statistical analyses and privacy risk assessments. Data utility was assessed by computing two heart failure risk scores on the protected data sets. We observed only minimal deviations to scores from the original data set. Additionally, we performed a re-identification risk analysis and found only minor residual risks for common types of privacy threats. Conclusion We could demonstrate that anonymization and synthetization methods protect privacy while retaining data utility for heart failure risk assessment. Both approaches and a combination thereof introduce only minimal deviations from the original data set over all features. While data synthesis techniques produce any number of new records, data anonymization techniques offer more formal privacy guarantees. Consequently, data synthesis on anonymized data further enhances privacy protection with little impacting data utility. We share all generated data sets with the scientific community through a use and access agreement.

A Functional Linear Regression for High‐Resolution 3D Faces

ABSTRACTMany scientific disciplines are faced with the challenge of extracting meaningful information from large, complex and highly structured datasets. A significant portion of contemporary statistical research is dedicated to developing tools for handling such data. This paper introduces a functional linear regression model specifically designed for 3D facial shapes, which are viewed as manifolds. We propose a comprehensive framework that includes converting 3D facial data into functional objects, employing a functional principal component analysis method and utilising a function‐on‐scalar regression model. This framework facilitates computation for high‐dimensional data and is employed to investigate how individual traits, such as age and genetic ancestry, impact the diversity of human facial features.

Explainable early detection of Alzheimer’s disease using ROIs and an ensemble of 138 3D vision transformers

Early detection and accurate diagnosis of brain morphological abnormalities are essential for the effective management and treatment of Alzheimer’s disease (AD) and mild cognitive impairment (MCI). Structural magnetic resonance imaging (MRI) is a powerful support tool to aid in disease diagnosis and prediction. In this research study, we present an innovative approach to predict Alzheimer’s disease (AD) and mild cognitive impairment (MCI) using MRI data, which integrates regional interest (ROI)-based methodology and deep learning within a comprehensible framework. The proposed method involves dividing the brain into 138 predetermined sections based on anatomical information. Next, we apply three-dimensional vision transformers (3D-ViTs) to each ROI individually, harnessing the power of deep learning. To improve prediction accuracy, we employ a deep belief network (DBN) as an ensemble learning model. Evaluating our approach on the baseline structural MRI dataset obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) cohort, and comparing it against five other competing models, we demonstrate its performance across four binary classification tasks and a three-class classification test (AD vs MCI vs CN (Cognitively Normal)). The proposed system outperforms existing models and provides interpretable insights into the brain regions that significantly contribute to solving each classification problem. Our findings align with the existing body of literature and hold promise for guiding future research directions in this domain.

Multiscale prediction model for autogenous shrinkage of early-age concrete incorporating high volume fly ash

High volume fly ash concrete (HVFAC) manufacture is an effective strategy to mitigate the autogenous shrinkage, achieve green production and low cost in building engineering. Nevertheless, a reasonable and robust multiscale prediction model of autogenous shrinkage of HVFAC and relative input parameters remains lacking. Thus, the objective of present study is to make up the missing gap. The research offered the input parameters dataset of microscopic pore structure, internal relative humidity (IRH) and validation dataset of autogenous shrinkage of cement pates and concretes with a water to binder ratio of 0.30/0.40, with fly ash in cement mass replacing-levels of 0, 30 %, 50 % and 70 %. The multiscale autogenous shrinkage prediction model was established by three parts of bulk modulus homogenization, capillary pressure calculation and composite models. The needed volumetric ratios of multi-phases were obtained by VCCTL software; bulk modulus was homogenized by Mori-Tanaka method, the proposed prediction models of IRH and LF-NMR-based microscopic pores characterization were used for capillary pressure calculation at cement paste level; composite models for autogenous shrinkage transformation from paste level to concrete level. The results show that HVFAC is capable of improved shrinkage-related pore structure, IRH and low autogenous shrinkage, which backs up the application potential of HVFAC in building engineering, and the input parameters dataset is confirmed to be reasonable. The proposed multiscale model was validated to be robust via comparing with test results and references dataset. The achievement of this study bear significance for HVFAC design and application in early-age cracking prevention in building engineering.

Interpretable neural network classification model using first-order logic rules

Over the past decade, the field of neural networks has made significant strides, particularly in deep learning. However, their limited interpretability has constrained their application in certain critical domains, drawing widespread criticism. Researchers have proposed various methods for explaining neural networks to address this challenge. This paper focuses on rule-based explanations for neural network classification problems. We propose IRCnet, a scalable classification model based on first-order logic rules. IRCnet consists of layers for learning conjunction and disjunction rules, utilizing binary logic activation functions to enhance interpretability. The model is initially trained using a continuous-weight version, which is later binarized to produce a discrete-weight version. During training, we innovatively employed gradient approximation method to handle the non-differentiable weight binarization function, thereby enabling the training of split matrices used for binarization. Finally, Conjunctive Normal Form (CNF) or Disjunctive Normal Form (DNF) rules are extracted from the model’s discrete-weight version. Experimental results indicate that our model achieves the highest or near-highest performance across various classification metrics in multiple structured datasets while demonstrating significant scalability. It effectively balances classification accuracy with the complexity of the generated rules.

The influence of CNN architecture, image size and quality to object detection model on histological specimens

The problem of improving the quality of a convolutional neural network (CNN) in the case of searching for objects on histological scans has been around for a long time and comes down, first of all, to choosing the correct CNN scheme and preparing a dataset of stable quality. The operation of the object detection algorithm is influenced by many factors, including image quality, image size, and the search for the object itself. Search for modern studies demonstrating the influence of various image characteristics on the selection of training programs and the choice of CNN design on the quality of the created model. As research basis, the literature for the past 5years devoted to data preprocessing, methodologies, requirements for images included in datasets, creating images for CNN models, and structure selection issues were analyzed. At the time of the study, the requirements for image sizes were not formulated and there is no data on the sizes of objects to the image sizes for optimizing the model. In addition, the problem of choosing a neural network scheme is not transparent and algorithmic. In most cases, researchers use the structures they developed or used themselves, without explaining the reasons or criteria for their choice, or comparing alternatives. All these issues complicate the process of developing CNN models for digital image processing. Current research presented and provided a brief overview of studies on preparing images for a dataset and possible choice of CNN structure.

PRFold-TNN: Protein Fold Recognition With an Ensemble Feature Selection Method Using PageRank Algorithm Based on Transformer.

Understanding the tertiary structures of proteins is of great benefit to function in many aspects of human life. Protein fold recognition is a vital and salient means to know protein structure. Until now, researchers have successively proposed a variety of methods to realize protein fold recognition, but the novel and effective computational method is still needed to handle this problem with the continuous updating of protein structure databases. In this study, we develop a new protein structure dataset named AT and propose the PRFold-TNN model for protein fold recognition. Firstly, different types of feature extraction methods including AAC, HMM, HMM-Bigram and ACC are selected to extract corresponding features for protein sequences. Then an ensemble feature selection method based on PageRank algorithm integrating various tree-based algorithms is used to screen the fusion features. Ultimately, the classifier based on the Transformer model achieves the final prediction. Experiments show that the prediction accuracy is 86.27% on the AT dataset and 88.91% on the independent test set, indicating that the model can demonstrate superior performance and generalization ability in the problem of protein fold recognition. Furthermore, we also carry out research on the DD, EDD and TG benchmark datasets, and make them achieve prediction accuracy of 88.41%, 97.91% and 95.16%, which are at least 3.0%, 0.8% and 2.5% higher than those of the state-of-the-art methods. It can be concluded that the PRFold-TNN model is more prominent.

The Dragoman Renaissance Research Project in Library/DH Linked Data Partnerships

The Dragomans Renaissance Research Platform represents a long-standing collaboration between the authors. The project explores the role of dragomans (diplomatic interpreter-translators) in mediating relations between the Ottoman Empire and its European neighbours from ca. 1550 to ca. 1730. The project leverages surrogates and transcriptions from multiple archival and secondary sources to produce a structured data set that can be queried using SPARQL. The Dragomans use case illuminates the potential of library-partnered DH projects, like other unique collections, to become machine-readable research data sets made available through generalized, sustainable architectures that also suit researcher needs for UI-driven query and display.

Aviation-BERT-NER: Named Entity Recognition for Aviation Safety Reports

This work introduces Aviation-BERT-NER, a Named Entity Recognition (NER) system tailored for aviation safety reports, building on the Aviation-BERT base model developed at the Georgia Institute of Technology’s Aerospace Systems Design Laboratory. This system integrates aviation domain-specific data, including aircraft types, manufacturers, quantities, and aviation terminology, to identify named entities critical for aviation safety analysis. A key innovation of Aviation-BERT-NER is its template-based approach to fine-tuning, which utilizes structured datasets to generate synthetic training data that mirror the complexity of real-world aviation safety reports. This method significantly improves the model’s generalizability and adaptability, enabling rapid updates and customization to meet evolving domain-specific requirements. The development process involved careful data preparation, including the synthesis of entity types and the generation of labeled datasets through template filling. Testing on real-world narratives from the National Transportation Safety Board (NTSB) database highlighted Aviation-BERT-NER’s robustness, with a precision of 95.34%, recall of 94.62%, and F1 score of 94.78% when evaluated over 50 manually annotated (BIO tagged) paragraphs. This work addresses a critical gap in English language NER models for aviation safety, promising substantial improvements in the analysis and understanding of aviation safety reports.