Statistical Modeling Methods Research Articles

A critical review of multi-output support vector regression

Single-output regression is a widely used statistical modeling method to predict an output based on one or more features of a datapoint. If a dataset has multiple outputs, they can be predicted independently from each other although this disregards potential correlations and thus may negatively affect the predictive performance. Therefore, multi-output regression methods predict multiple outputs simultaneously. One way to approach single-output regression is by using methods based on support vectors such as support vector regression (SVR) or least-squares SVR (LS-SVR). Based on these two, previous works have devised multi-output support vector regression methods. In this review, we introduce a unified notation to summarize the single-output support vector regression methods SVR and LS-SVR as well as state-of-the-art multi-output support vector regression methods. Furthermore, we implemented a workflow for subject- and record-wise bootstrapping and nested cross-validation experiments, which we used for an exhaustive evaluation of all single- and multi-output support vector regression methods on synthetic and non-synthetic datasets. Although the reviewed papers claim that their multi-output methods improve regression performance, we find that none of them outperform both single-output methods SVR and LS-SVR for various reasons. Due to these results, we reflected about the general concept of support vector regression and then concluded that support vector regression methods do not appear to be suitable for the task of multi-output regression.

Analysis of key genes in the development of lung cancer

Lung cancer is the cancer with the highest incidence rate. Due to the insidious early symptoms, more than 80% of lung cancer patients are diagnosed at an advanced stage. If certain key genes that have a main effect on oncogenic ability can be identified, researchers will be able to use precisely targeted therapies to predict and treat the cancer. In order to identify the key genes that play an important role in the development of lung cancer, this study collected gene expression data respectively from lung cancer patients and paraneoplastic samples, and used statistical modeling methods such as Pearson correlation analysis and significance test to screen out typical genes that have a high degree of variability in cancer and paraneoplastic samples and a high correlation with the survival time of patients. The study obtained 1487 genes that were significantly up-regulated in cancer samples and 447 significantly down-regulated genes. Additionally, these key genes were enriched for pathways relying on the GO and KEGG databases to obtain the common functions of these genes and to explore the biological mechanisms of lung cancer development.

A comparative analysis of the methods of evaluating dependability using Big Data of complex system operation

Aim. The paper aims to research the methods of evaluating dependability indicators based on the analysis of Big Data obtained in the course of monitoring the operation of technical systems and their components. A comparative analysis is performed of the efficiency of robust statistical methods of evaluating the dependability indicators of complex technical systems based on operation data.Methods. The paper uses methods of mathematical statistics, specifically robust methods of evaluating the translation parameter of a noisy sample and numerical methods of statistical modeling. The authors consider five methods of evaluating the translation parameter: sample mean as a nonrobust method used for comparison; sample median as the simplest robust method of evaluating the translation parameter; two-stage evaluation procedure with truncation of outliers according to the three sigma rule; two-stage evaluation procedure with truncation of outliers using Tukey’s box-and-whisker plot; Huber’s robust method. The comparative analysis of the methods of evaluating system dependability indicators was conducted by means of statistical modeling in the R statistical computation package. Five distribution laws for generating an element’s time-to-failure and recovery time samples were considered: exponential distribution, Weibull distribution, log-normal distribution, gamma distribution and uniform distribution.Results. Statistical analysis of Big Data associated with the operation of technical systems is complicated by the heterogeneity and noisiness of such data, as well as the presence of errors and outliers of varied nature. That is primarily due to the varied loads and operating conditions of each object. Herein this problem is examined as regards the problem of evaluating the dependability indicators of a structurally complex monotonic system with independent element recovery. The paper examines methods of rejecting anomalous data and robust evaluation of the sample position parameter and performs a comparative analysis of the efficiency of such methods for various distribution laws. It is shown that robust methods of evaluation enable significantly higher accuracy as compared to the standard sample mean. The two-stage procedure based on the truncation of outliers and Tukey’s box-and-whisker plot proved to be the most efficient.Conclusions. The paper’s findings allow improving the accuracy of evaluation of dependability indicators based on complex technical system operation data. They can be used in Big Data processing and complex system dependability theory.

APPLICATION OF THE STATISTICAL TESTING METHOD FOR SIMULATION MODELLING OF COMPLEX SYSTEMS

Problem statement. The study of the multicomponent systems’ operation is possible using statistical simulation models. In contrast to traditional mathematical modeling, in order to build the specified type of models, it is not necessary to obtain a mathematical formalization of the relationship between parameters in one form or another. The main condition is the ability of the model to reproduce the phenomena being modeled, while preserving their logical and temporal sequence and physical meaning. One of the most common methods of statistical simulation modeling is the method of statistical tests - the Monte Carlo method. The basis of this method is the repeated use of a random number generator to simulate the dynamics of processes occurring in the system. Based on the obtained iterations, statistical criteria for evaluating the obtained results are calculated, which makes it possible to conduct a preliminary analysis of the physical processes under investigation and draw conclusions about the relationship between the parameters included in the model. Thus, taking into account the general principles of these building simulation models, it is advisable to use them to study a number of applied materials science problems, for example, to describe the influence of external factors on the structural state of the material. The purpose of the article. Application of statistical simulation modeling by the method of statistical tests (Monte Carlo method) for the study of physical processes that occur in complex systems. Conclusion. Using the Monte Carlo method, a simulation model of the relationship between the technological modes of welding and the parameters of the structural state of low-carbon low-alloy steel 09Г2С was obtained. The obtained simulation model is presented in the form of a matrix containing the results of statistical tests. The analysis of the obtained statistical indicators made it possible to carry out a preliminary analysis of the relationship between technological modes of welding and the corresponding structural state of low-carbon low-alloy steel 09Г2С. Preliminary evaluation of the obtained simulation model was carried out by analyzing active constraints of the type of inequalities. The conducted set of studies proved the obtained data adequacy.

Factors having the most significant impact on university students’ level of financial literacy: Forecasting based on cognitive methodology

Introduction. The authors explore the problem of increasing financial literacy in modern university students based on an analysis of the results of a sociological survey and cognitive modeling tools. The purpose of the study is to identify a set of interrelated factors that have the most significant impact on the level of financial literacy and predict the features of the development of the system. Materials and Methods. During the research, the survey method, mathematical and statistical methods for processing empirical data, expert assessment and computer modeling were used. The methodology of cognitive modeling, founded by R. Axelrod, was used as a methodological basis of the study. In order to identify factors that have the most significant impact on the level of financial literacy, a survey was conducted involving 195 Omsk university students. Based on the results of the survey and expert assessment, a cognitive model of the level of financial literacy was developed and a series of computer experiments were conducted. Results. During the research process, the highest priority topics in the field of financial education were revealed; a set of factors influencing the level of financial literacy was identified, connections between them were determined and an expert assessment of their intensity was carried out; a cognitive model of the level of financial literacy in the form of a weighted directed graph has been developed; a series of computer experiments was conducted and the impact on the target factor of influences of varying intensity exerted on the associated control factors such as, the ability to plan a personal budget, etc. was predicted. Conclusions. The study showed the social significance of increasing the level of financial literacy in university students. The results of the computer modeling made it possible to predict to what extent influences made on control factors can determine the level of financial literacy and to draw conclusions about the highest priority topics. The results obtained can serve as a scientific basis for designing financial literacy courses in modern universities.

Global root trait underrepresentation in Paleotropical communities: a qualitative analysis

Many ecosystem processes revolve around plant roots and the rhizosphere, but root trait knowledge has generally lagged behind aboveground plant traits from leaves and stems. In particular, root trait representation of Paleotropical species and tropical communities in modern trait databases remains poor, constraining our understanding of belowground processes in these regions. Therefore, we sought to examine the current state of root trait representation and associated topics in the Paleotropics to identify gaps and biases in the existing literature. Using an exhaustive literature scan, we compiled a list of publications that contained various belowground trait information. Subsequently, we utilized a statistical topic modeling method to analyze abstracts of publications in order to identify topics studied alongside root trait documentation. Finally, we consolidated trends in root trait coverage and topics across five geographical regions and four time periods to illustrate shifts in literature and knowledge of ecological processes revolving around roots. Root trait representation was heavily biased toward root biomass but other aspects of root systems such as physiology, architecture and anatomical traits remain underrepresented. We categorized 23 unique topics around root trait literature across eight categories: ecosystem productivity and biomass stocks, plant functional traits, resource availability, ecosystem processes and dynamics, mycorrhizal colonization, edaphic processes, seedling experiments, and global change and variation. Traits and topic coverage were unequally distributed across the Paletropics and exhibited a notable shift in focus from resource limitation and mycorrhizae research to root trait variation at large spatial and temporal scales over the last 50 years of root trait literature. Given these trends and the heterogenous effects global change exerts on the tropics at a regional scale, we provided several recommendations to facilitate inter-study comparisons of traits to advance the field’s understanding of belowground ecosystem processes in pantropical communities.

Three-dimensional lumbar spine generation using variational autoencoder.

The disease analysis of the lumbar spine often requires a large number of three-dimensional (3D) models. Currently, there is a lack of 3D model of the lumbar spine for research, especially for the diseases such as scoliosis where it is difficult to collect sufficient data in a short period of time. To solve this problem, we develop an end-to-end network based on 3D variational autoencoder for randomly generating 3D lumbar spine model. In this network, the dual path encoder structure is used to fit two individual variables, i.e., mean and variance. Spatial coordinate attention modules are added to the encoder to improve the learning ability of the network to the 3D spatial structure of the lumbar spine. To enhance the power of the network to reconstruct the lumbar spine, a regularization loss is added to constrain the distribution loss. Additionally, Gaussian noise layers are added to the decoder to improve the authenticity and diversity of generated model. The experiments were conducted on the data of the entire lumbar spine and the individual lumbar vertebra, respectively. The results showed that the voxel intersection over union was 0.588 and 0.684, the voxel Dice coefficient was 0.739 and 0.811, the average surface distance was 0.807 and 1.189, and the Hausdorff distance was 2.615 and 3.710, for the entire lumbar spine and individual lumbar vertebra, respectively. The developed approach is comparable to the most commonly used model generation method of statistical shape model (SSM) in both visual and objective indicators, while the developed approach does not require the landmarks that is needed in the SSM method. Therefore, this fully automatic method can be easily used for population-based modeling of the lumbar spine which has the potential to be a powerful clinical tool.

Using coherent processing algorithms for direction finding of UAV acoustic signals

Small UAVs are often used in a group, since their signals are highly correlated, their resolution is reduced when using non-coherent processing.
 The article analyzes the well-known methods of coherent processing of acoustic signals in order to increase the resolution in the direction finding of signals with a high degree of correlation. Obtaining qualitative indicators of the analyzed algorithms was carried out by the method of statistical computer modeling in the Matlab environment.
 Based on the simulation results, it is shown that coherent signal processing methods are the most stable in conditions of low signal-to-noise ratios, while non-coherent ones show the best results in the region of high signal-to-noise ratio, while coherent algorithms can potentially distinguish more targets. the WAVES coherent algorithm performs better in the high signal-to-noise ratio region, but loses to the CSSM algorithm in the low signal-to-noise region.
 To increase the efficiency of coherent processing of multipath signals, it applies spatial filtering of the inputsignal.

Burden ofischemic stroke inmainland China andTaiwan province from 1990 to 2019: with forecast forthe next 11 years.

Ischemic stroke is featured with high incidence, mortality, and disability. The aim of this study is to use Global Burden of Disease database to describe and compare the burden of ischemic stroke in mainland China and Taiwan province and to further predict the expected changes in the next 11 years using statistical modeling methods. Information on ischemic stroke incidence and mortality in China (mainland and Taiwan province) during 1990-2019 was obtained from the Global Burden of Disease database to analyze the effects of region, gender, and age on the incidence and mortality of ischemic stroke in China. The autoregressive integrated moving average model was used to predict the age-standardized incidence rate and age-standardized mortality rate of ischemic stroke in males and females in mainland China and Taiwan province in the next 11 years. The period from 1990 to 2019 witnessed an overall upward trend in the number of incidence and deaths in mainland China and Taiwan province. In 2019, there were nearly 2.87 million ischemic incidence cases with stroke in mainland China, with more female patients than male in the age group of over 60 years. Among the nearly 1.03 million deaths, the death toll of men under the age of 85 years was higher than that of women, while in Taiwan province, the number of incidence was 28 771, with more female patients of all ages than male. Among the 6788 deaths, the death toll of men under the age of 80 years was higher than that of women. In 2019, the age group with the highest number of patients in the two regions was 65-69 years, while the highest number of deaths was found in people aged 85 years and above. As our autoregressive integrated moving average model predicted, the age-standardized incidence rate value of ischemic stroke is expected to be 163.23/100 000 persons in mainland China by 2030, which would continue to increase, while the age-standardized mortality rate value of ischemic stroke is expected to be 16.41/100 000 persons in Taiwan province by 2030, which showed a decreasing trend. Disease burden of ischemic stroke is still increasing in mainland China and Taiwan province, and health resources should be deployed to implement effective prevention and control strategies, taking into account region, gender, and age.

Prediction of Rapid Early Progression and Survival Risk with Pre-Radiation MRI in WHO Grade 4 Glioma Patients.

Recent clinical research describes a subset of glioblastoma patients that exhibit REP prior to the start of radiation therapy. Current literature has thus far described this population using clinicopathologic features. To our knowledge, this study is the first to investigate the potential of conventional radiomics, sophisticated multi-resolution fractal texture features, and different molecular features (MGMT, IDH mutations) as a diagnostic and prognostic tool for prediction of REP from non-REP cases using computational and statistical modeling methods. The radiation-planning T1 post-contrast (T1C) MRI sequences of 70 patients are analyzed. An ensemble method with 5-fold cross-validation over 1000 iterations offers an AUC of 0.793 ± 0.082 for REP versus non-REP classification. In addition, copula-based modeling under dependent censoring (where a subset of the patients may not be followed up with until death) identifies significant features (p-value < 0.05) for survival probability and prognostic grouping of patient cases. The prediction of survival for the patients' cohort produces a precision of 0.881 ± 0.056. The prognostic index (PI) calculated using the fused features shows that 84.62% of REP cases fall under the bad prognostic group, suggesting the potential of fused features for predicting a higher percentage of REP cases. The experimental results further show that multi-resolution fractal texture features perform better than conventional radiomics features for prediction of REP and survival outcomes.

Spatio-temporal variability of seawater mixing in the coastal aquifers based on hydrogeochemical fingerprinting and statistical modeling

This study discusses monitoring and characterization of spatial and temporal variability to comprehend groundwater salinization based on hydrogeochemical fingerprinting and statistical modeling in the coastal belt of Digha-Shankarpur-Tajpur-Mandarmani, West Bengal, India. An integrated study of hydrogeochemical, bulk magnetic susceptibility, multivariate statistical, and geochemical modeling methods is implemented. The major cationic and anionic concentrations in groundwater are in the order Na+ > Ca2+ > Mg2+ > K+ and Cl− > HCO3− > SO4− > NO3− > F− respectively. The major water types are dominated by (Ca2+ – HCO3−) followed (Ca2+ - Mg2+ - Cl−), (Ca2+ - Na+ - HCO3−), (Na+ - HCO3−), and (Na+ - Cl). The results showed that the groundwater quality continuously declined steadily from pre-monsoon 2020 to pre-monsoon 2022. The deterioration of groundwater is due to an interplay of multiple factors, i.e., water-rock interaction, including ion-exchange, seawater mixing, and anthropogenic actions. Furthermore, it is also found that the regions showing higher seawater mixing index and oversaturated with carbonate minerals are also areas where groundwater is unsuitable for irrigation. The findings are beneficial in assisting local communities and legislators in designing appropriate management and mitigation techniques to arrest seawater intrusion in coastal regions.

A Smart Farm DNN Survival Model Considering Tomato Farm Effect

Recently, smart farming research based on artificial intelligence (AI) has been widely applied in the field of agriculture to improve crop cultivation and management. Predicting the harvest time (time-to-harvest) of crops is important in smart farming to solve problems such as planning the production schedule of crops and optimizing the yield and quality. This helps farmers plan their labor and resources more efficiently. In this paper, our concern is to predict the time-to-harvest (i.e., survival time) of tomatoes on a smart farm. For this, it is first necessary to develop a deep learning modeling approach that takes into account the farm effect on the tomato plants, as each farm has multiple tomato plant subjects and outcomes on the same farm can be correlated. In this paper, we propose deep neural network (DNN) survival models to account for the farm effect as a fixed effect using one-hot encoding. The tomato data used in our study were collected on a weekly basis using the Internet of Things (IoT). We compare the predictive performance of our proposed method with that of existing DNN and statistical survival modeling methods. The results show that our proposed DNN method outperforms the existing methods in terms of the root mean squared error (RMSE), concordance index (C-index), and Brier score.

A novel optimization framework for minimizing the surface roughness while increasing the material processing rate in the SLM process of 316L stainless steel

Purpose316 L stainless steel alloy is potentially the most used material in the selective laser melting (SLM) process because of its versatility and broad fields of applications (e.g. medical devices, tooling, automotive, etc.). That is why producing fully functional parts through optimal printing configuration is still a key issue to be addressed. This paper aims to present an entirely new framework for simultaneously reducing surface roughness (SR) while increasing the material processing rate in the SLM process of 316L stainless steel, keeping fundamental mechanical properties within their allowable range.Design/methodology/approachConsidering the nonlinear relationship between the printing parameters and features analyzed in the entire experimental space, machine learning and statistical modeling methods were defined to describe the behavior of the selected variables in the as-built conditions. First, the Box–Behnken design was adopted and corresponding experimental planning was conducted to measure the required variables. Second, the relationship between the laser power, scanning speed, hatch distance, layer thickness and selected responses was modeled using empirical methods. Subsequently, three heuristic algorithms (nonsorting genetic algorithm, multi-objective particle swarm optimization and cross-entropy method) were used and compared to search for the Pareto solutions of the formulated multi-objective problem.FindingsA minimum SR value of approximately 12.83 μm and a maximum material processing rate of 2.35 mm3/s were achieved. Finally, some verification experiments recommended by the decision-making system implemented strongly confirmed the reliability of the proposed optimization methodology by providing the ultimate part qualities and their mechanical properties nearly identical to those defined in the literature, with only approximately 10% of error at the maximum.Originality/valueTo the best of the authors’ knowledge, this is the first study dealing with an entirely different and more comprehensive approach for optimizing the 316 L SLM process, embedding it in a unique framework of mechanical and surface properties and material processing rate.

Система непрерывного мониторинга технического состояния и оперативной диагностики паровой турбины

Power equipment deterioration occurs during operation, it causes loss in reliability and efficiency, unscheduled shutdowns, and accidents. Now predictive analytics is one of the promising directions in the field of power engineering which allows to control and analyze the technical condition of power equipment. The problems of localization of deviation of technological parameters and detection of anomaly in the operation of power equipment are consistently solved in the framework of predictive analytics. The problems of deviations localization and anomalies detection are solved by the methods of statistical modeling and the classification algorithms respectively. However, for steam turbines the localization of deviation and the detection of anomalies having slow-flowing character are a difficult problem. Therefore, the issue of development of a method for continuous monitoring of technical condition and maintenance diagnostics based on a mathematical model of the steam turbine section flow characteristics is worth noticing. The method allows us to consider the effect of changes in the open flow area of the individual sections of a steam turbine on the pressure distribution over the steam flow path. The model of a steam turbine has been developed within the matrix formalization methodology. The solution of the system of linear and nonlinear equations is carried out by methods of computational mathematics. The solution of the optimization problems of the steam flow path diagnostics is carried out by methods of mathematical programming. A mathematical model of the cogeneration steam turbine Т-250/305-23.5-DB and a method for continuous condition monitoring and maintenance diagnostics of steam turbine have been developed. It allows us to localize deviation and detect anomaly by recovery of the open flow area of the individual sections of a steam turbine based on the pressure distribution over the steam flow path. The results of the statistical analysis prove that a mathematical model of the cogeneration steam turbine Т-250/305-23.5-DB has been recognized adequate. The method for continuous condition monitoring and maintenance diagnostics of steam turbine has demonstrated consistency of the obtained results and ability to solve diagnostic problems in practice. The developed model and method can be used as a module in the development of a software package for predictive analytics of power equipment.

Experimental Aging and Lifetime Prediction in Grid Applications for Large-Format Commercial Li-Ion Batteries

Due to the growth of electric vehicle and stationary energy storage markets, the production and use of lithium-ion batteries has grown exponentially in recent years. For many of these applications, large-format lithium-ion batteries are being utilized, as large cells have less inactive material relative to their energy capacity and require fewer electrical connections to assemble into packs. The economics of these battery systems is highly dependent on cell lifetime, requiring accelerated aging tests to characterize cell degradation. However, testing of large-format lithium-ion batteries is time consuming and requires high current channels and large testing chambers, so not many reports on the lifetime of commercial, large-format lithium-ion batteries exist. Here, accelerated aging test data from four commercial large-format lithium-ion batteries is reported. These batteries span both NMC-Gr and LFP-Gr cell chemistries, are from a variety of manufacturers, have pouch and prismatic formats, and show a range of cell designs with varying power capabilities.Accelerated aging test results are analyzed to examine both cell performance, in terms of efficiency and thermal response under load, as well as cell lifetime. Cell thermal response is characterized by measuring temperature during cycle aging, which is used to calculate a normalized thermal resistance value that may help estimate both cell cooling needs or to help extrapolate aging test results to different thermal environments. Cell lifetime is evaluated qualitatively, considering the average calendar and cycle life across a range of conditions, as well as quantitatively, using statistical modeling and machine-learning methods to identify predictive aging models from the accelerated aging data. These predictive aging models are then used to investigate cell sensitivities to stressors, such as cycling temperature, voltage window, and C-rate, as well as to predict cell lifetime in various stationary storage applications.Results from this work show that cell lifetime and sensitivity to aging conditions varies substantially across the cell types, necessitating testing for each specific cell to make quantitative lifetime predictions. That being said, all commercial cells tested here are predicted to reach at least 10-year lifetimes for stationary storage applications. Based on the aging test results and modeling, some cells are expected to be relatively insensitive to temperature and use-case, making them suited for simple use cases with little or no thermal management and simple controls, while the lifetime of other cells could be extended to 20+ years if operated with thermal management and degradation-aware controls. Figure 1

Identification of road traffic crashes hotspots on an intercity expressway in India using geospatial techniques

Ascertaining the underlying pattern of road traffic crashes (RTCs) and identifying hotspots is essential for improving safety on the road network. Researchers have employed various statistical modelling and spatial methods to predict crash frequency and identify their hotspots on the road network. In India, the road network length has been increasing, especially the expressway network length. The increase in the network length has also increased RTCs. Hence, it is essential to assess the crash pattern and identify hotspots on the intercity expressways in India. This study aims to identify the fatal crash hotspots on the selected intercity expressway using geospatial methods. First, in this study, hotspot sections were identified using ordinary kriging (OK) and, kernel density estimation (KDE), network kernel density estimation (NKDE) methods. Next, the employed techniques were compared to know their predictive effectiveness in identifying the hotspots. The study used the fatal crash data from August 2012 to October 2018 for the selected 165 km intercity expressway. Outcomes of the geospatial methods revealed some of the common hotspots are identified by both methods. The comparative analysis indicated that the NKDE method is more effective in identifying the hotspots in smaller segments than the other two methods. Consequently, this research's outcomes would facilitate intercity expressway-owning agencies to select a practical and readily applicable hotspot identification methodology in LMICs.

Variation Aware Optimization vs Classical Margin Optimization of Superconducting Electronics

We have developed a Superconducting Electronics Circuit Design Optimizer in conjunction with a Statistical Process Variation Josephson Junction Model and formal method of Logic Verification as a means to optimize circuits and mitigate their susceptibility to factors of yield loss. In this paper we will contrast the Yield improvement provided by direct yield optimization against classical methods of margin centering.

A Real-Time Vehicle Speed Prediction Method Based on a Lightweight Informer Driven by Big Temporal Data

At present, the design of modern vehicles requires improving driving performance while meeting emission standards, leading to increasingly complex power systems. In autonomous driving systems, accurate, real-time vehicle speed prediction is one of the key factors in achieving automated driving. Accurate prediction and optimal control based on future vehicle speeds are key strategies for dealing with ever-changing and complex actual driving environments. However, predicting driver behavior is uncertain and may be influenced by the surrounding driving environment, such as weather and road conditions. To overcome these limitations, we propose a real-time vehicle speed prediction method based on a lightweight deep learning model driven by big temporal data. Firstly, the temporal data collected by automotive sensors are decomposed into a feature matrix through empirical mode decomposition (EMD). Then, an informer model based on the attention mechanism is designed to extract key information for learning and prediction. During the iterative training process of the informer, redundant parameters are removed through importance measurement criteria to achieve real-time inference. Finally, experimental results demonstrate that the proposed method achieves superior speed prediction performance through comparing it with state-of-the-art statistical modelling methods and deep learning models. Tests on edge computing devices also confirmed that the designed model can meet the requirements of actual tasks.

Patterns of aeroallergen sensitization in asthma patients identified by latent class analysis: A cross-sectional study in China.

This cross-sectional study aimed to identify latent sensitization profiles of asthma patients in mainland China, unveiling the association between regional differences and sensitization patterns. 1056 asthma participants from 10 medical centers divided into eastern and western cohorts were clustered into four individual sensitization patterns, respectively, by using an unsupervised statistical modeling method, latent class analysis (LCA), based on the levels of 12 aeroallergens specific IgE reactivities. Moreover, differences in clinical characteristics and environmental exposures were compared in different sensitization patterns. Four distinct sensitization patterns in the two cohorts were defined as follows, respectively. Eastern cohort: Class 1: "High weed pollen and house dust mites (HDMs) sensitization" (8.87%), Class 2: "HDMs dominated sensitization" (38.38%), Class 3: "High HDMs and animal dander sensitization" (6.95%), Class 4: "Low/no aeroallergen sensitization" (45.80%). Western cohort: Class 1: "High weed pollen sensitization" (26.14%), Class 2: "High multi-pollen sensitization" (15.02%), Class 3: "HDMs-dominated sensitization" (10.33%), Class 4: "Low/no aeroallergen sensitization" (48.51%). Of note, the significant statistical difference in age, asthma control test score (ACT) and comorbidities were observed within or between different sensitization patterns. Exposure factors in different sensitization patterns were pointed out. Asthmatic patients with distinct sensitization patterns were clustered and identified through the LCA method, disclosing the relationship between sensitization profiles of multiple aeroallergens and geographical differences, providing novel insights and potential strategies for atopic disease monitoring, management and prevention in clinical practice.

Linguistic and statistical analysis of the lexical ‘Langue-Parole’ dichotomy in a restricted domain

Development of new digital methods for analyzing the ‘Langue-Parole’ dichotomy is one of the most sought-after, but least researched problems of modern theoretical and applied linguistics. This determines the relevance of this study, the purpose of which is to develop a methodology for the automated linguastatistical analysis of a domain-related lexical layer in the context of the ‘Langue-Parole’ dichotomy and to apply the methodology to the Russian-language domain “Research on athlete integrative physiology” (RAIP). The study was conducted on the material of the Russian-language corpus including 56 RAIP domain texts of 300,000 wordforms in total published over the 2013-2020 period in the scientific journals “People. Sport. Medicine” (formerly “SUSU Bulletin. Series “Education, Healthcare, Physical Culture”), “Theory and Practice of Physical Culture”, etc. The key methodological approach is the ontological analysis of corpus data using statistical and linguistic modeling methods. The domain-specific language and speech are modeled by the corresponding lexicon and corpus, while the ‘Langue-Parole’ lexical dichotomy is represented by the values of the linguistic-statistical concept verbalization parameters of the domain concepts in the lexicon and corpus. The computational parameters include the indices of lexical diversity, structural complexity, conceptual syncretism, lexical structural complexity vs. conceptual syncretism correlation, and syncretical concept junction when verbalized in the corpus. The main results of the study are: 1) а methodology for analyzing domain-specific lexical dichotomy ‘Langue-Parole’, which can be ported to other domains and national languages; 2) the RAIP domain-related resources, including language-independent ontology, conceptually annotated Russian corpus, onto-lexicon, linguistic-statistical parameter values of the lexical ‘Langue-Parole’ dichotomy; and 3) tools that automate certain stages of the study.