Section Classification Research Articles

Structural performance and section classification of austenitic stainless steel welded I-section flexural members

Structural performance and section classification of austenitic stainless steel welded I-section flexural members

Utilization of Robson 10-group caesarean section classification among midwives and obstetricians in Nyahururu County Referral Hospital, Kenya

Utilization of Robson 10-group caesarean section classification among midwives and obstetricians in Nyahururu County Referral Hospital, Kenya

Local-global buckling interaction in steel I-beams—A European design proposal for the case of fire

This paper presents a comprehensive review of the developments leading to the proposal of European fire design rules for steel beams with thin-walled I-sections. Thin-walled beams are favoured in steel construction due to their structural efficiency, however they are prone to buckle when subjected to in-plane loads, phenomenon which requires a thorough investigation. The focus lies on addressing the interaction between local and global buckling in these members. This study discusses modifications to the Effective Width Method at elevated temperatures, aiming to rectify underestimation of section capacity on certain cross-sections. Additionally, an overview of lateral-torsional buckling resistance is provided, being a new Effective Section Factor proposed to account for its interaction with local buckling. The paper revisits the European fire design proposal, comparing the improvements of the existing Eurocode 3 Part 1–2 (EN1993-1-2:2005) relative to the recent second generation of Part 1–2 of Eurocode 3 (FprEN1993-1-2:2023) through an extensive numerical study. Limitations and future research directions are also discussed, such as dependency on the section classification and broadening the application scope of these rules to higher steel grades.

Clinical information extraction for lower-resource languages and domains with few-shot learning using pretrained language models and prompting

Abstract A vast amount of clinical data are still stored in unstructured text. Automatic extraction of medical information from these data poses several challenges: high costs of clinical expertise, restricted computational resources, strict privacy regulations, and limited interpretability of model predictions. Recent domain adaptation and prompting methods using lightweight masked language models showed promising results with minimal training data and allow for application of well-established interpretability methods. We are first to present a systematic evaluation of advanced domain-adaptation and prompting methods in a lower-resource medical domain task, performing multi-class section classification on German doctor’s letters. We evaluate a variety of models, model sizes (further-pre)training and task settings, and conduct extensive class-wise evaluations supported by Shapley values to validate the quality of small-scale training data and to ensure interpretability of model predictions. We show that in few-shot learning scenarios, a lightweight, domain-adapted pretrained language model, prompted with just 20 shots per section class, outperforms a traditional classification model, by increasing accuracy from $48.6\%$ to $79.1\%$ . By using Shapley values for model selection and training data optimization, we could further increase accuracy up to $84.3\%$ . Our analyses reveal that pretraining of masked language models on general-language data is important to support successful domain-transfer to medical language, so that further-pretraining of general-language models on domain-specific documents can outperform models pretrained on domain-specific data only. Our evaluations show that applying prompting based on general-language pretrained masked language models combined with further-pretraining on medical-domain data achieves significant improvements in accuracy beyond traditional models with minimal training data. Further performance improvements and interpretability of results can be achieved, using interpretability methods such as Shapley values. Our findings highlight the feasibility of deploying powerful machine learning methods in clinical settings and can serve as a process-oriented guideline for lower-resource languages and domains such as clinical information extraction projects.

EHR-HGCN: An Enhanced Hybrid Approach for Text Classification Using Heterogeneous Graph Convolutional Networks in Electronic Health Records.

Text classification is a central part of natural language processing, with important applications in understanding the knowledge behind biomedical texts including electronic health records (EHR). In this article, we propose a novel heterogeneous graph convolutional network method for classifying EHR texts. Our method, called EHR-HGCN, is able to combine context-sensitive word and sentence embeddings with structural sentence-level and word-level relation information to perform text classification. EHR-HGCN reframes EHR text classification as a graph classification task to better capture structural information about the document using a heterogeneous graph. To mine contextual information from a document, EHR-HGCN first applies a bidirectional recurrent neural network (BiRNN) on word embeddings obtained via Global Vectors for word representation (GloVe) to obtain context-sensitive word-level and sentence-level embeddings. To mine structural relationships from the document, EHR-HGCN then constructs a heterogeneous graph over the word and sentence embeddings, where sentence-word and word-word relationships are represented by graph edges. Finally, a heterogeneous graph convolutional neural network is used to classify documents by their graph representation. We evaluate EHR-HGCN on a variety of standard text classification benchmarks and find that EHR-HGCN has higher accuracy and F1-score than other representative machine learning and deep learning methods. We also apply EHR-HGCN to the MedLit benchmark and find it performs with high accuracy and F1-score on the task of section classification in EHR texts. Our ablation experiments show that the heterogeneous graph construction and heterogeneous graph convolutional network are critical to the performance of EHR-HGCN.

A kurtosis-based parameter for classifying elliptical hollow sections under bending

A novel section parameter termed normalized excess kurtosis of a section, inspired by the same parameter used in probability distributions, is introduced to characterize the rotation capacity of a hollow section under bending. The proposed normalization accounts for variations in yield stress and Young’s modulus too. A linear relationship is observed between rotation capacity and the normalized excess kurtosis of circular hollow sections (CHS), rectangular/square hollow sections (RHS) and elliptical/oval hollow sections (EHS), under constant and linearly varying moment conditions, based on experimental and finite element model (FEM) data. It is found that, the rotation capacity variation of hollow sections is better explained by using normalized excess kurtosis than by the conventional section slenderness. The gradient of the above linear relationship varies with the section type and provides an estimate of the rotation capacity of a hollow section with a known shape and aspect ratio. It also provides insights into the section classification of EHS sections, with aspect ratios closer to unity (0.67 < a/b < 1.5) being suitable for an equivalent diameter approach and those with aspect ratios further from unity (a/b < 0.5 and a/b > 2.0) for an equivalent RHS approach. The difference between the moment and rotation capacities of linearly varying moment conditions (3-point bending) and constant moment conditions (4-point bending) is also elucidated.

Hysteretic performance of thin-walled Q690 high-strength steel H-section beam-columns bent about the weak-axis

High-strength steel has broad application prospects in steel structures as its weight to strength ratio is low, but the reduced ductility limits its application in seismic design. In order to evaluate the application potential of high-strength steel in light steel structures, four Q690 high-strength steel beam–columns with large width to thickness ratios were tested under cyclic loading about the weak axis. The failure mode, ultimate strength, hysteretic curve, stiffness degradation, ductility and energy dissipation capacity of the tested specimens are discussed. Subsequently, a finite element analysis model is established, and an extensive parametric study is carried out using the verified finite element model. The test results show that the failure mode is the local buckling of column base. The hysteretic behavior of the specimens is mainly affected by the axial load ratio and the flange width to thickness ratio. The results of the parametric study show that the extent of development of plasticity in the cross-section is related to the axial load ratio. The calculated ultimate strength of the beam–column shows that Eurocode 3 is conservative for the classification of the beam–column as having a Class 4 cross-section, members with Class 4 cross-sections should in fact be classified as having Class 3 cross-sections. Accordingly, a section classification for a high-strength steel beam–column subjected to weak axis bending is discussed.

Behaviour of glass fiber reinforced polymer (GFRP) structural profile columns under axial compression

Abstract Glass fiber-reinforced plastic (GFRP) structural profiles in the construction industry are a promising alternative to conventional building materials due to their high strength-to-weight ratio, high tensile strength, insulation properties, chemical resistance, fatigue properties, and lower maintenance cost. This study aims to evaluate the compression behaviour of semi-compact and slender glass fiber-reinforced plastic I-sections. Short, intermediate, and long columns of both slender and semi-compact I-sections were subjected to axial compression, and the experimental capacity was compared to available theoretical results. It was found that the experimental capacity of short, semi-compact, and long columns were 27 %, 49 %, and 40 % lower than the theoretical capacity of semi-compact I-sections. Short slender sections had an ultimate experimental capacity 55 % greater than the theoretical results. However, lower ultimate capacities were achieved for intermediate and long columns in the case of slender sections when compared to the theoretical capacity. Slender sections were prone to both global and local buckling, whereas semi-compact sections failed by global buckling alone. The study also concludes the need for proper section classification of glass fiber-reinforced plastic sections to consider buckling characteristics. The existing theoretical equations to estimate the load-carrying capacity was found to be overly conservative; hence it is necessary to conduct numerical and parametric studies to develop equations that are more in agreement with the experimental results.

Improving model transferability for clinical note section classification models using continued pretraining.

The classification of clinical note sections is a critical step before doing more fine-grained natural language processing tasks such as social determinants of health extraction and temporal information extraction. Often, clinical note section classification models that achieve high accuracy for 1 institution experience a large drop of accuracy when transferred to another institution. The objective of this study is to develop methods that classify clinical note sections under the SOAP ("Subjective," "Object," "Assessment," and "Plan") framework with improved transferability. We trained the baseline models by fine-tuning BERT-based models, and enhanced their transferability with continued pretraining, including domain-adaptive pretraining and task-adaptive pretraining. We added in-domain annotated samples during fine-tuning and observed model performance over a varying number of annotated sample size. Finally, we quantified the impact of continued pretraining in equivalence of the number of in-domain annotated samples added. We found continued pretraining improved models only when combined with in-domain annotated samples, improving the F1 score from 0.756 to 0.808, averaged across 3 datasets. This improvement was equivalent to adding 35 in-domain annotated samples. Although considered a straightforward task when performing in-domain, section classification is still a considerably difficult task when performing cross-domain, even using highly sophisticated neural network-based methods. Continued pretraining improved model transferability for cross-domain clinical note section classification in the presence of a small amount of in-domain labeled samples.

Risk section classification of tunnel settlement based on land-use development simulation and uncertainty analysis

• Propose a method to identify the settlement risk section along metro lines. • Extend Artificial Neural Network–Cellular Automata on uncertainty analysis. • Present a detailed case study for verification. “Metro economy” has led to intensive land development along the metro lines. However, engineering activities associated with land development inevitably disturb the service environment, causing severe settlement of adjacent metro tunnel structures in soft soil areas. As a reference point for pre-treating this type of tunnel settlement, a method is proposed to classify settlement risk sections along metro lines based on a land-use development simulation and corresponding uncertainty analysis. First, the land-use development along the metro line was simulated by Artificial Neural Network–Cellular Automata (ANN-CA). Second, the land-use development process was considered a random event rather than a deterministic prediction as in a typical ANN-CA, with its probability quantified based on the cells’ conversion probability. The classification of the surrounding land-use development probability was used to allocate the settlement risk sections of metro lines. This method was applied to the Han–You section of the Nanjing Metro Line 2. The predicted settlement risk sections corresponded suitably with the actual settlement troughs, demonstrating the effectiveness of this method. Thus, this method provides a novel consideration for the pre-treatment of metro tunnel settlement from the perspective of interactions between the metro line and surrounding land development.

Clustering of Asphalt Pavement Maintenance Sections Based on 3D Ground-Penetrating Radar and Principal Component Techniques

Asphalt pavement maintenance section classification is an important prerequisite for accurately determining asphalt pavement maintenance needs and formulating accurate maintenance plans. This paper introduces the three-dimensional (3D) ground-penetrating radar (GPR) pavement internal crack rate index on the basis of an original road surface performance data matrix, and the dimensionality of the road section classification data matrix was reduced through the principal component technique. An analysis of variance was used to compare the significance of the differences in the results for road section classification using different clustering methods and different clustering data and to investigate the influence of the clustering method, principal component technique and crack rate index on the maintenance road section classification results. The results showed that the principal component technique could reduce the dimensionality of the data matrix by 33% and retain more than 84% of the information. There was a genetic relationship between the clustering data and the technical characteristics of the classified sub-sections, and the internal crack rate was important for the characterisation of internal defects in asphalt pavement sub-sections and the determination of maintenance needs. The results of section classification varied considerably between clustering methods, and the choice of clustering method had a relationship to the pavement maintenance objectives. The dynamic clustering method combined with principal component analysis could significantly improve the significance of the differences in the clustering results, effectively improving the division of maintenance sections.

Rectangular filled composite members made from high strength materials: Behavior and design of columns and beams

The AISC Specification limits the use of steel and concrete materials in filled composite members to concrete strength (fc′) less than or equal to 69 MPa and steel yield strength (Fy) less than or equal to 517 MPa. This paper presents the results of comprehensive analytical investigations conducted using benchmarked numerical models to evaluate the behavior of rectangular filled composite members made from one or both materials exceeding the AISC Specification strength limits. The numerical models are developed and analyzed using the displacement-based formulation of the fiber-based finite element in OpenSees. Prior research has developed effective (phenomenological) stress–strain curves for steel and concrete materials that implicitly account for the effects of steel yielding, local buckling, concrete cracking, crushing and confinement, and the interaction between the two materials in the composite cross-section. This paper first benchmarks the modeling approach and then uses the benchmarked model to conduct analytical parametric studies covering a wide range of parameters, including: member slenderness, section slenderness, section aspect ratio, steel yield stress and concrete compressive strength. The results from the parametric studies are used to evaluate behavior and develop new design equations based on nominal stress distributions over the composite section. The nominal steel stress in compression (Fcr) is expressed as a continuous function of the section slenderness and yield strength, thus eliminating the current AISC Specification’s need for section classification (compact, noncompact and slender). The new design equations are recommended for relaxing the material strength limitations in the next version of the AISC Specification.

Cesarean Section Classification Using Machine Learning With Feature Selection, Data Balancing, and Explainability

Cesarean Section Classification Using Machine Learning With Feature Selection, Data Balancing, and Explainability

Comparison of Section Classification Procedure of Different Codes

The cross-section classification is essential to identify local buckling in steel sections. Local buckling restricts the section from attaining its maximum moment capacity. Cross section classification is based on the behavior of the element under direct and bending compressive stresses. Limit on width to thickness ratio of elements in a cross section is the governing criterion for the section classification in different national codes. In the case of seismic design, the post yield ductility is an important parameter; therefore, it is particularly essential to prevent local buckling. Design codes of different countries provide different limiting values for the section classification. The present study discusses the approach of various national codes towards cross section classification. The approach of Eurocode 3, AISC 341, AISC 360, BS 5950 and IS 800 on issues like section classification of members subjected to axial and flexural compression, sections with elements of different classes are explained and their salient features are compared with each other. It is observed from the study that, amongst the codes considered, AISC provides the most stringent limits for seismically compact section, whereas, for other class of sections Eurocode 3 limits are more stringent.

Classifying caesarean section to understand rising rates among Palestinian refugees: results from 290,047 electronic medical records across five settings

BackgroundRising caesarean-section rates worldwide are driven by non-medically indicated caesarean-sections. A systematic review concluded that the ten-group classification system (Robson) is the most appropriate for assessing drivers of caesarean deliveries. Evidence on the drivers of caesarean-section rates from conflict-affected settings is scarce. This study examines caesareans-section rates among Palestinian refugees by seven-group classification, compares to WHO guidelines, and to rates in the host settings, and estimates the costs of high rates.MethodsElectronic medical records of 290,047 Palestinian refugee women using UNRWA’s (United Nations Relief and Works Agency for Palestine Refugees in the Near East) antenatal service from 2017–2020 in five settings (Jordan, Lebanon, Syria, West Bank, Gaza) were used. We modified Robson criteria to compare rates within each group with WHO guidelines. The host setting data were extracted from publicly available reports. Data on costs came from UNRWA’s accounts.FindingsPalestinian refugees in Gaza had the lowest caesarean-section rates (22%), followed by those residing in Jordan (28%), West Bank (30%), Lebanon (50%) and Syria (64%). The seven groups caesarean section classification showed women with previous caesarean-sections contributed the most to overall rates. Caesarean-section rates were substantially higher than the WHO guidelines, and excess caesarean-sections (2017–2020) were modelled to cost up to 6.8 million USD. We documented a steady increase in caesarean-section rates in all five settings for refugee and host communities; refugee rates paralleled or were below those in their host country.InterpretationCaesarean-section rates exceed recommended guidance within most groups. The high rates in the nulliparous groups will drive future increases as they become multiparous women with a previous caesarean-section and in turn, face high caesarean rates. Our analysis helps suggest targeted and tailored strategies to reduce caesarean-section rates in priority groups (among low-risk women) organized by those aimed at national governments, and UNRWA, and those aimed at health-care providers.

Ultimate biaxial bending resistance of H-section steel members under different loading paths

In order to explore the ultimate biaxial bending performance of H-section steel members under different loading conditions, we conduct a comprehensive parametric study on H-section members with different axial force ratios, web and flange width-to-thickness ratios, and loading angles. Material and geometric nonlinearity are considered, and previous laboratory test results validate the finite element models. A criterion for determining the ultimate state of biaxial bending members is proposed based on the elastic–plastic stability theory. The ultimate biaxial capacity of the members subjected to monotonic biaxial bending and axial force is analyzed, and biaxial moment ultimate interaction curves are obtained by least square fitting. Furthermore, the influences of four different hysteretic protocols on the hysteretic behavior of the members are investigated, and the applicability of the monotonic biaxial moment limit interaction curve to the cyclic loading model is verified. A design equation for predicting the ultimate capacity of the members is proposed, considering the hardening effect of the material and the interaction effect of the plate. The proposed approach can predict the ultimate capacity of H-section biaxial bending members. The method is not limited by section classification and loading protocol and has universal applicability.

Fire resistance design of austenitic SUS316 stainless columns subjected to axial compression

To explore the effect of cold-working on the properties of austenitic stainless steel, a material mechanical property test of austenitic SUS316 stainless steel coupons extracted from cold-formed RHS columns at elevated temperatures was conducted. Based on the fire test of 5 austenitic stainless SHS columns subjected to axial compression, a refined finite element model was established. A single parametric analysis of different parameters, cross-sectional area, height-thickness ratio, initial overall imperfection, eccentricity, slenderness ratio, aspect ratio and load ratio was conducted, mainly investigating the effects on the critical temperature and maximum axial displacement. As the effect of the load ratio was the most significant, a multiparameter analysis of the load ratio with other parameters was further conducted. Presently, the design methods in EN 1993–1-2 for carbon steel columns are also applied to stainless steel columns, where section classification is needed. Based on numerous parametric analyses, the design methods in EN 1993–1-2 were assessed and modified. EN 1993–1-2 underestimated the bearing capacity of stainless steel SHS and RHS columns, and the modified method improved the utilization rate of stainless steel. A new method for evaluating the bearing capacity of stainless SHS and RHS columns using the maximum axial displacement was proposed, for which the section classification was not needed. As a preliminary study, the new method was only applicable to stainless SHS and RHS columns with load ratios between 0.2 and 0.6, which was also the most commonly applied scope in engineering.

Compressive strength of high strength concrete-encased composite columns with noncompact and slender steel angles

Concrete-encased steel angle (CEA) column is an encased composite column where structural angles are used for longitudinal and transverse reinforcements. In this study, the compressive behavior of CEA columns was investigated, focusing on the use of high-strength concrete and noncompact and slender steel angles. Ten CEA column specimens (concrete strength fc' = 33 and 66 MPa and steel yield strength Fy = 454–552 MPa) were tested under compressive loading. The section classification of steel angles and the spacing of transverse ties were considered as test variables. The tests showed that the cover concrete spalled off at peak loads and local buckling occurred in steel angles. Due to the use of high-strength concrete, the compressive strength (i.e., peak loads) was increased but the post-peak degradation behavior became steeper. The peak and residual strengths of CEA columns were examined focusing on the effects of cover failure, confinement by transverse ties, and local buckling of steel angles. Based on the results, empirical equations to estimate the peak and residual strengths were proposed and the validity was assessed by comparisons with the test results.

Analysis of Cardiac Ultrasound Images of Critically Ill Patients Using Deep Learning.

Cardiovascular disease remains a substantial cause of morbidity and mortality in the developed world and is becoming an increasingly important cause of death in developing countries too. While current cardiovascular treatments can assist to reduce the risk of this disease, a large number of patients still retain a high risk of experiencing a life-threatening cardiovascular event. Thus, the advent of new treatments methods capable of reducing this residual risk remains an important healthcare objective. This paper proposes a deep learning-based method for section recognition of cardiac ultrasound images of critically ill cardiac patients. A convolution neural network (CNN) is used to classify the standard ultrasound video data. The ultrasound video data is parsed into a static image, and InceptionV3 and ResNet50 networks are used to classify eight ultrasound static sections, and the ResNet50 with better classification accuracy is selected as the standard network for classification. The correlation between the ultrasound video data frames is used to construct the ResNet50 + LSTM model. Next, the time-series features of the two-dimensional image sequence are extracted and the classification of the ultrasound section video data is realized. Experimental results show that the proposed cardiac ultrasound image recognition model has good performance and can meet the requirements of clinical section classification accuracy.

Automatic Detection of Transmission Line Dangerous Points Based on Point Cloud Tower Section Classification Mode

In the actual laser power inspection data processing process, the classification error of point cloud is inevitable. Some classification errors often need to be located with the help of dangerous points which is detected by mistake. Therefore, in the process of laser power inspection point cloud data processing, the detection of Dangerous points often needs multiple calculations. However, the dangerous point detection is often based on the whole line, and involves the data calculation of hundreds of tower section at the same time. The data magnitude is large and the calculation speed is slow, which seriously affects the efficiency of point cloud category correction. Aiming at the above problems, this paper divides the point cloud into tower segments, transforms the calculation unit of dangerous point detection from a whole line to a single tower segment, and records the modification of the point category in the tower segment. In the process of correcting the classification error and repeatedly calculating the dangerous points, only the tower segment with the modified category is recalculated. The practical application shows that this method can effectively speed up the data processing efficiency of laser power inspection.