Analysis Algorithm Research Articles

PopCluster: A Population Genetics Model-Based Toolset for Simulating, Inferring and Visualising Individual Admixture and Population Structure.

In this computer note I introduce software, PopCluster, that implements a new likelihood method for unsupervised population structure analysis from marker data. To infer a coarse population structure, it assumes the mixture model and adopts a simulated annealing algorithm to make a maximum likelihood clustering analysis, partitioning the sampled individuals into a predefined number of clusters. To deduce a fine population structure, it further assumes the admixture model and employs an expectation maximisation algorithm to estimate individual admixture proportions. PopCluster has many features. First, it is one of just a couple of model-based methods that can handle both biallelic and multiallelic markers in the same framework. Second, it is the first population structure analysis method that uses both Message Passing Interface (MPI) and openMP to exploit multiple CPUs with both shared and distributed memories and has the capacity to handle genomic data with millions of individuals and millions of loci. Third, the algorithms for both mixture and admixture analyses are fast, rendering PopCluster favourably in computational efficiency over previous methods in analysing genomic data. Fourth, PopCluster is built for Windows, Linux and Mac platforms, and its Windows version has an integrated GUI that can conveniently visualise analysis results and facilitate data input. Fifth, its Windows version has a built-in simulation module designed to simulate genotype data under admixture, hybridization or migration models. PopCluster provides a valuable toolset for researchers to simulate, infer and visualise individual admixture and population genetic structure, hybridization and migration using marker data.

Algorithm for Detecting Reference Points on a Digital Electrocardiogram in Real Time

Relevance. The use of digital electrocardiographs and cardiac monitors with built-in algorithms for automatic processing, analysis and interpretation of electrocardiograms allows the doctor to effectively diagnose cardiac arrhythmias. It is known that in order to provide emergency care to a patient, the duration of arrhythmia diagnostics should not exceed several tens of seconds, which requires the emergence of new algorithms for detecting informative features indicating arrhythmia, operating in real time. The need to introduce new and effective technologies for diagnosing cardiovascular diseases is also reflected in public health development programmes.Research goal. Development and quality indicators analysis of the algorithm for reference points detection on a digital electrocardiogram, bearing informative signs for the procedure of arteries diagnosis.The methods used. The study is based on an analysis of existing approaches to the problem of reference points detection on digital electrocardiogram, as well as conducting a test of the proposed algorithms by mathematical modelling methods. The quality indicators of the algorithms defined in accordance with the principles of signal detection theory and diagnostic testing, at the junction of which the task of electrocardiogram reference point detection is located. The proposed algorithm was tested on materials of MIT-BIH Arrhythmia Database, which is widely used for verification and validation of real-time digital electrocardiogram signal processing algorithms.The results. The study proposes an algorithm for detecting reference points on a digital electrocardiogram that carry informative features for the arrhythmia diagnostic procedure. The proposed algorithm is based on digital signal filtering using a decision rule based on a three-step two-threshold principle of pre-processed electrocardiogram signal values comparison on a sliding window. An experiment on the materials of the open verified MIT-BIH Arrhythmia DB showed that the quality of the proposed algorithm for detecting reference points is higher than that of the algorithms used in modern digital electrocardiographs and cardiac monitors. The proposed algorithm based on digital signal filtering and the three-step two-threshold decision rule have elements of scientific novelty.The significance. The results of this work can be used in the development of digital heart rate monitors, cardiac devices and for automatic processing, analysis and real-time computer-assisted digital electrocardiogram signal interpretation.

Optimizing Quality Control: A Comprehensive Analysis of Computer Vision Methods for Assessing Vegetables and Fruits

Efficient quality control in the agriculture sector, particularly regarding the inspection of vegetables and fruits, stands as a critical necessity in today's health-focused industry. Conventional fruit grading methods, ill-suited for large-scale production, demand an automated, non-invasive, and economically feasible substitute. Computer vision emerges as a promising avenue, leveraging image analysis and machine learning algorithms to evaluate the quality of produce. The convergence of computer vision and image processing technologies in contemporary agriculture has brought about a substantial transformation in quality assessment methodologies. This paper conducts an in-depth exploration of the amalgamation of computer vision and image processing techniques for the evaluation of agricultural produce quality. Through a comprehensive review, this scientific analysis investigates the integration of computer vision and image processing techniques in agricultural quality assessment. It scrutinizes key studies, their practical implementations, outcomes, and the research voids they reveal. Technological progressions within the agricultural domain have the potential to amplify productivity and curtail the circulation of flawed or substandard products. Moreover, this study deliberates on the forthcoming trends in computer vision technology applications, accentuating their prospective influence on the vegetables and fruits industry.

Computational Simulation Study-Based Formulation Development and Characterization of MethylprednisoloneLoaded Nanoparticles Containing Chitosan and Pectin to Treat Nocturnal Asthma.

Nocturnal asthma (NA) is a high-prevalence disease that causes severe respiratory issues, leading to death from early midnight to early morning. In this research, nanoparticulate drug delivery system of methylprednisolone (MP) was developed using chitosan (CH) and pectin (PEC). MP is a synthetic corticosteroid medication widely used for its potent anti-inflammatory activity. Computational simulation study (AI-based blend analysis algorithm) was used to identify a better-mixing polymer with MP. MP nanoparticles were formulated by the ionic gelation method with the combination of CH and PEC. To modify the drug release properties, the formed beads were coated with chitosan succinate (CSSC). The morphological characteristics of the beads were determined by SEM analysis. The X-ray radiographic imaging study was used to observe the intactness of MP beads. Histopathological studies were also carried out to find out the toxicity of the beads in the organs of rats. Pectin and chitosan polymers were selected based on the computational simulation study. SEM analysis revealed that the beads had a spherical shape with a rough outer surface. CSSC-coated beads achieved sustained drug release for up to 24 h. X-ray imaging demonstrated the stability of the beads in acidic pH conditions. In vivo pharmacokinetic studies showed that CSSC-coated beads were more stable in the gastrointestinal tract (GIT) than PEC-CH beads and the pure drug. Histological evaluation confirmed that the beads are nontoxic and safe for use in rats. Based on the findings, it was concluded that CSSC-coated beads of MP exhibited superior release properties, making them suitable for a chronomodulated drug delivery system.

Application of Neuroinformatics in Alzheimer’s Disease Research

Alzheimer’s disease (AD) is a neurodegenerative disorder characterized by the gradual decline in cognitive function, and its complex pathological mechanisms remain incompletely understood. As an interdisciplinary field that integrates multimodal data, neuroinformatics has played a crucial role in Alzheimer’s disease research in recent years. This review systematically explores the application of neuroinformatics in Alzheimer’s disease research, covering key databases and resources (e.g., ADNI and GAAIN), imaging technologies (e.g., MRI, fMRI, EEG, PET), and data analysis algorithms (e.g., support vector machines, convolutional neural networks). Neuroinformatics techniques have provided new pathways for early diagnosis, risk prediction, and personalized treatment of Alzheimer’s disease by integrating imaging, genomic, behavioral, and clinical data to reveal changes in brain function and structure in AD patients. In the future, with the strengthening of big data platforms and international collaborations, neuroinformatics will further promote the development of precision medicine in Alzheimer’s disease and provide stronger support for new treatment strategies.

High-performance breast cancer diagnosis method using hybrid feature selection method.

One of the primary causes of the women death is breast cancer. Accurate and early breast cancer diagnosis plays an essential role in its treatment. Computer Aided Diagnosis (CAD) system can be used to help doctors in the diagnosis process. This study presents an efficient method to performance improvement of the breast cancer diagnosis CAD system using thermal images. The research strategy in the proposed CAD system is using efficient algorithms in feature extraction and classification phases, and new efficient feature selection algorithm. In the feature extraction phase, the Segmentation Fractal Texture Analysis (SFTA) algorithm that is a texture analysis algorithm is used.This algorithm utilizes two-threshold binary decomposition. In the feature selection phase, the developed feature selection algorithm, which is hybrid of binary grey wolf optimization algorithm and firefly optimization algorithm, is applied to extracted features. Then, the kNN, SVM, and DTree classification techniques are applied to check whether the selected features are efficiently discriminated the group successfully with minimal misclassifications. The DMR database is utilized for performance evaluation of the proposed method. The results indicate that the obtained accuracy, specificity, sensitivity, and MCC are 97, 96, 98, and 94.17 %, respectively. The developed breast cancer diagnosis method has advantages compared to other breast cancer diagnosis using thermal images.

Advanced integration of IoT and AI algorithms for comprehensive smart meter data analysis in smart grids

Abstract Smart grids utilize advanced communication, information, and control technologies to build a more friendly power supply system. However, the development of technology leads to the increasing complexity of smart meter (SM) data in the power grid, and analyzing SM data becomes difficult. Therefore, in order to realize the effective analysis and forecasting of users’ electricity consumption behavior, an SM data analysis technique integrating the Internet of Things and artificial intelligence is proposed, which contains three major techniques: meter reading systems, load clustering, and load forecasting, and at last the technique is validated. Finally, the technology is validated. The experimental results showed that the average value of the Davis-Balding index for the proposed method combining recursive graphs and convolutional autoencoders was 1.87, which was significantly lower than the average value of 2.35 for other methods. The average value of the Dunn index was 0.086, which was higher than the 0.065 for the comparison method. In the load forecasting model, the model that employs multi-task learning and a gated cycle unit demonstrates superior performance across all comparison schemes, with the lowest average absolute percentage error of 12.90%. This is significantly lower than the comparison model’s 15.35%, which highlights the enhanced efficacy of the proposed model. The study realizes the effective analysis and prediction of consumers’ electricity consumption behavior and provides new ideas and solutions for the optimization and management of smart grids.

Machine learning-based identification of histone deacetylase-associated prognostic factors and prognostic modeling for low-grade glioma.

Low-grade glioma (LGG) is a slow-growing but invasive tumor that affects brain function. Histone deacetylases (HDACs) play a critical role in gene regulation and tumor progression. This study aims to develop a prognostic model based on HDAC-related genes to aid in risk stratification and predict therapeutic responses. Expression data from The Cancer Genome Atlas (TCGA) and Chinese Glioma Genome Atlas (CGGA) were analyzed to identify an optimal HDAC-related risk signature from 73 genes using 10 machine learning algorithms. Patients were stratified into high- and low-risk groups based on the median risk score. Prognostic accuracy was evaluated using Kaplan-Meier survival analysis and receiver operating characteristic (ROC) curves. Functional enrichment analyses, including Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Set Enrichment Analysis (GSEA), were performed to explore pathways linked to the gene signature. Immune infiltration and tumor microenvironment characteristics were assessed using Single Sample Gene Set Enrichment Analysis (ssGSEA) and ESTIMATE algorithm. SubMap was applied to predict responsiveness to immune checkpoint inhibitors, and chemotherapeutic sensitivity was analyzed via the Genomics of Drug Sensitivity in Cancer (GDSC) database. A prognostic model consisting of four HDAC-related genes-SP140, BAZ1B, SP100, and SIRT1-was identified. This signature displayed strong prognostic accuracy, achieving a C-index of 0.945. Individuals with LGG were systematically divided into high-risk and low-risk cohorts based on the median risk value, enabling more precise risk stratification. The survival prognosis was significantly worse in the high-risk cohort compared to the low-risk group, highlighting distinct survival trajectories. Notably, the two cohorts exhibited marked shifts in immune checkpoint gene transcriptional profiles and immune cell infiltration maps, underscoring fundamental biological differences that contribute to these differing prognoses. We developed an HDAC-related four-gene prognostic model that correlates with survival, immune landscape, and therapeutic response in LGG patients. This model may guide personalized treatment strategies and improve prognostic accuracy, warranting further validation in clinical settings.

Reduced-order estimation and optimisation of claw-shaped throttling notch via CFD analysis, surrogate models, and CCMO algorithm

Reduced-order estimation and optimisation of claw-shaped throttling notch via CFD analysis, surrogate models, and CCMO algorithm

Understanding consumer perception and acceptance of AI art through eye tracking and Bidirectional Encoder Representations from Transformers-based sentiment analysis

This study investigates public perception and acceptance of AI-generated art using an integrated system that merges eye-tracking methodologies with advanced bidirectional encoder representations from transformers (BERT)-based sentiment analysis. Eye-tracking methods systematically document the visual trajectories and fixation spots of consumers viewing AI-generated artworks, elucidating the inherent relationship between visual activity and perception. Thereafter, the BERT-based sentiment analysis algorithm extracts emotional responses and aesthetic assessments from numerous internet reviews, offering a robust instrument for evaluating public approval and aesthetic perception. The findings indicate that consumer perception of AI-generated art is markedly affected by visual attention behavior, whereas sentiment analysis uncovers substantial disparities in aesthetic assessments. This paper introduces enhancements to the BERT model via domain-specific pre-training and hyper- parameter optimization utilizing deep Gaussian processes and dynamic Bayesian optimization, resulting in substantial increases in classification accuracy and resilience. This study thoroughly examines the underlying mechanisms of public perception and assessment of AI-generated art, assesses the potential of these techniques for practical application in art creation and evaluation, and offers a novel perspective and scientific foundation for future research and application of AI art.

AI-Driven Real-Time Feedback System for Enhanced Student Support: Leveraging Sentiment Analysis and Machine Learning Algorithms

The rapid evolution of educational technologies has led to a shift toward personalized and adaptive learning experiences. A critical component of such systems is the ability to provide timely and relevant feedback to students. This paper presents an AI-driven real-time feedback system designed to enhance student support through the integration of sentiment analysis and machine learning algorithms. The system leverages sentiment analysis to gauge the emotional tone of student interactions, such as forum posts, assignment submissions, and feedback. Machine learning algorithms, including decision trees, support vector machines (SVM), and deep learning models, are used to analyze and predict student engagement, performance, and emotional states. By combining both cognitive and emotional insights, the system delivers personalized, context-sensitive feedback that helps students overcome learning challenges and improve academic outcomes. The effectiveness of the system is evaluated using multiple datasets, showing significant improvements in student engagement, satisfaction, and performance.

Decoding Informal Settlements in Core Urban Areas of Karachi: Leveraging Machine Learning Algorithms for Classification and Analysis

Decoding Informal Settlements in Core Urban Areas of Karachi: Leveraging Machine Learning Algorithms for Classification and Analysis

K-hyperparameter tuning in high-dimensional genomics using joint optimization of deep differential evolutionary algorithm and unsupervised transfer learning from intelligent GenoUMAP embeddings

AbstractK-hyperparameter optimization in high-dimensional genomics remains a critical challenge, impacting the quality of clustering. Improved quality of clustering can enhance models for predicting patient outcomes and identifying personalized treatment plans. Subsequently, these enhanced models can facilitate the discovery of biomarkers, which can be essential for early diagnosis, prognosis, and treatment response in cancer research. Our paper addresses this challenge through a four-fold approach. Firstly, we empirically evaluate the k-hyperparameter optimization algorithms in genomics analysis using a correlation based feature selection method and a stratified k-fold cross-validation strategy. Secondly, we evaluate the performance of the best optimization algorithm in the first step using a variety of the dimensionality reduction methods applied for reducing the hyperparameter search spaces in genomics. Building on the two, we propose a novel algorithm for this optimization problem in the third step, employing a joint optimization of Deep-Differential-Evolutionary Algorithm and Unsupervised Transfer Learning from Intelligent GenoUMAP (Uniform Manifold Approximation and Projection). Finally, we compare it with the existing algorithms and validate its effectiveness. Our approach leverages UMAP pre-trained special autoencoder and integrates a deep-differential-evolutionary algorithm in tuning k. These choices are based on empirical analysis results. The novel algorithm balances population size for exploration and exploitation, helping to find diverse solutions and the global optimum. The learning rate balances iterations and convergence speed, leading to stable convergence towards the global optimum. UMAP’s superior performance, demonstrated by short whiskers and higher median values in the comparative analysis, informs its choice for training the special autoencoder in the new algorithm. The algorithm enhances clustering by balancing reconstruction accuracy, local structure preservation, and cluster compactness. The comprehensive loss function optimizes clustering quality, promotes hyperparameter diversity, and facilitates effective knowledge transfer. This algorithm’s multi-objective joint optimization makes it effective in genomics data analysis. The validation on this algorithm on three genomic datasets demonstrates superior clustering scores. Additionally, the convergence plots indicate relatively smoother curves and an excellent fitness landscape. These findings hold significant promise for advancing cancer research and computational genomics at large.

Identification of amino acid metabolism‑related genes as diagnostic and prognostic biomarkers in sepsis through machine learning.

Previous research has highlighted the critical role of amino acid metabolism (AAM) in the pathophysiology of sepsis. The present study aimed to explore the potential diagnostic and prognostic value of AAM-related genes (AAMGs) in sepsis, as well as their underlying molecular mechanisms. Gene expression profiles from the Gene Expression Omnibus (GSE65682, GSE185263 and GSE154918 datasets) were analyzed. Based on weighted gene co-expression network analysis and machine learning algorithms, hub AAMGs were identified in the GSE65682 database. Subsequently, hub AAMGs were evaluated for their expression levels and diagnostic and prognostic significance in sepsis, as well as their interactions with regulatory pathways and role in immune cell infiltration. Additionally, trends in AAMG expression were validated using clinical samples, and their functions in sepsis were confirmed through an in vitro model. In total, four AAMGs were identified, two of which, methionine synthase (MTR) and methionine-R-isomerase 1 (MRI1), demonstrated significant differential expression in the GSE65682, GSE185263 and GSE154918 datasets, which was further validated using clinical samples. A diagnostic nomogram based on MTR and MRI1 expression demonstrated strong diagnostic effectiveness across the three aforementioned databases. Moreover, the expression of both genes were negatively correlated with sepsis prognosis and showed stratified prognostic capabilities. Newly identified pathways included KRAS and IL-2/STAT5 signaling. MTR and MRI1 negatively correlated with the infiltration of inflammatory cells, such as M1 macrophages and neutrophils, and positively correlated with anti-inflammatory cells, such as CD8+ T and dendritic cells. In vitro experiments further demonstrated that overexpression of MTR could mitigate the inhibition of cloning and proliferation induced by LPS and ATP in RAW 264.7 cells. These findings highlighted the potential of MTR and MRI1 as biomarkers for diagnosing and prognosticating sepsis, potentially acting through the regulation of methionine in the pathophysiology of this disease. The present study provided new insights into the role of AAM in the mechanisms underlying sepsis and in the potential development of future targeted therapies.

Pitfalls in the analysis conditions of canine brain perfusion computed tomography.

This study aimed to investigate the impact of selected analysis conditions on blood flow values and color maps in canine brain perfusion computed tomography (PCT) and to propose optimal analysis conditions. Dynamic computed tomography imaging was performed on six beagle dogs. Color maps were generated using a combination of analysis algorithms (box-modulation transfer function (Box-MTF) and singular value deconvolution plus (SVD+) methods), slice thicknesses (4.0 and 8.0 mm), analysis matrix sizes (512 × 512, 256 × 256, and 128 × 128), and noise reduction levels (strong and weak). Cerebral blood flow (CBF) and cerebral blood volume (CBV) were calculated for gray matter, white matter, basal ganglia, hippocampus, thalamus, and cerebellum in each map. CBF and CBV values obtained using SVD+ were significantly higher than those obtained using Box-MTF. Noise reduction was more effective with larger matrix sizes; however, excessive noise reduction led to the blurring of anatomical structures in the color map. Across all analysis algorithms, anatomical structures were challenging to visualize at 8.0 mm. For canine brain PCT, it is essential to choose a straightforward algorithm that remains unaffected by circulatory velocity or intracranial bone structure. Given the brain's size, the slice thickness should be minimal, noise reduction level should be suitable for the targeted area, and matrix size should be maximized.

DNEA: an R package for fast and versatile data-driven network analysis of metabolomics data

BackgroundMetabolomics is a high-throughput technology that measures small molecule metabolites in cells, tissues or biofluids. Analysis of metabolomics data is a multi-step process that involves data processing, quality control and normalization, followed by statistical and bioinformatics analysis. The latter step often involves pathway analysis to aid biological interpretation of the data. This approach is limited to endogenous metabolites that can be readily mapped to metabolic pathways. An alternative to pathway analysis that can be used for any classes of metabolites, including unknown compounds that are ubiquitous in untargeted metabolomics data, involves defining metabolite-metabolite interactions using experimental data. Our group has developed several network-based methods that use partial correlations of experimentally determined metabolite measurements. These were implemented in CorrelationCalculator and Filigree, two software tools for the analysis of metabolomics data we developed previously. The latter tool implements the Differential Network Enrichment Analysis (DNEA) algorithm. This analysis is useful for building differential networks from metabolomics data containing two experimental groups and identifying differentially enriched metabolic modules. While Filigree is a user-friendly tool, it has certain limitations when used for the analysis of large-scale metabolomics datasets.ResultsWe developed the DNEA R package for the data-driven network analysis of metabolomics data. We present the DNEA workflow and functionality, algorithm enhancements implemented with respect to the package’s predecessor, Filigree, and discuss best practices for analyses. We tested the performance of the DNEA R package and illustrated its features using publicly available metabolomics data from the environmental determinants of diabetes in the young. To our knowledge, this package is the only publicly available tool designed for the construction of biological networks and subsequent enrichment testing for datasets containing exogenous, secondary, and unknown compounds. This greatly expands the scope of traditional enrichment analysis tools that can be used to analyze a relatively small set of well-annotated metabolites.ConclusionsThe DNEA R package is a more flexible and powerful implementation of our previously published software tool, Filigree. The modular structure of the package, along with the parallel processing framework built into the most computationally extensive steps of the algorithm, make it a powerful tool for the analysis of large and complex metabolomics datasets.

Photovoltaic Short-Term Output Power Forecast Model Based on Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise–Kernel Principal Component Analysis–Long Short-Term Memory

To solve the problem of photovoltaic power prediction in areas with large climate changes, this article proposes a hybrid Long Short-Term Memory method to improve the prediction accuracy and noise resistance. It combines the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) and kernel principal component analysis (KPCA) algorithm. The ICEEMDAN algorithm reduces the instability of the environmental factor sequence. The KPCA algorithm reduces the input dimensions of the model. LSTM performs dynamic time modeling of the multivariate feature sequences to predict the output PV power. The adaptability of the ICEEMDAN-KPCA-LSTM model is assessed with datasets from a PV plant in west China and evaluated by root mean squared error (RMSE), mean absolute percentage error (MAPE), and R-squared metrics. Using 70% of the datasets for output PV power estimation, the results show a good performance, with an RMSE of 4.3715, MAPE of 8.9264%, and R-squared value of 89.973%. By comparing with other prediction models, the ICEEMDAN-KPCA-LSTM photovoltaic output power model outperforms other models.

Early diagnosis of Alzheimer’s disease and mild cognitive impairment using MRI analysis and machine learning algorithms

Early diagnosis of Alzheimer’s disease (AD) and mild cognitive impairment (MCI) is crucial to prevent their progression. In this study, we proposed the analysis of magnetic resonance imaging (MRI) based on features including; hippocampus (HC) area size, HC grayscale statistics and texture features (mean, standard deviation, skewness, kurtosis, contrast, correlation, energy, homogeneity, entropy), lateral ventricle (LV) area size, gray matter area size, white matter area size, cerebrospinal fluid area size, patient age, weight, and cognitive score. Five machine learning classifiers; K-nearest neighborhood (KNN), support vector machine (SVM), random forest (RF), decision tree (DT), and multi-layer perception (MLP) were used to distinguish between groups: cognitively normal (CN) vs AD, early MCI (EMCI) vs late MCI (LMCI), CN vs EMCI, CN vs LMCI, AD vs EMCI, and AD vs LMCI. Additionally, the correlation and dependence were calculated to examine the strength and direction of association between each extracted feature and each classification of the group. The average classification accuracies in 20 trials were 95% (SVM), 71.50% (RF), 82.58% (RF), 84.91% (SVM), 85.83% (RF), and 85.08% (RF), respectively, with the best accuracies being 100% (SVM, RF, and MLP), 83.33% (RF), 91.66% (RF), 95% (SVM, and MLP), 96.66% (RF), and 93.33% (DT). Cognitive scores, HC and LV area sizes, and HC texture features demonstrated significant potential for diagnosing AD and its subtypes for all groups. RF and SVM showed better performance in distinguishing between groups. These findings highlight the importance of using 2D-MRI to identify key features containing critical information for early diagnosis of AD.

Extracting Meaningful Information from Turkish Chemistry and Physics Texts with Machine Learning

This study emphasizes the importance of processing a dataset consisting of Turkish chemistry and physics texts created by us through artificial intelligence systems. A model is proposed to pave the way for artificial intelligence-based analyses and discoveries in the basic sciences of chemistry and physics. Chemistry and physics, the basic sciences, are critical in many industrial, medical, and environmental applications. However, significant data analysis is required to access and understand information in these areas. This study aims to demonstrate the effectiveness of machine learning methods in extracting meaningful information from Turkish chemistry and physics texts. For this purpose, the tokenization process is first performed, and then the features are extracted with Term frequency-inverse document frequency (TF-IDF) and Bag-of-Words (BOW) methods. The combined features are classified separately with Support Vector Machine (SVM), Naive Bayes (NB), Quadratic Discriminant Analysis (QDA), k-nearest Neighbors (KNN), Logistic Regression (LR), Random Forest (RF), and Gradient Boosting (GB) algorithms. According to the classification results, the best calculation time and the most successful accuracy rate are obtained with NB at 95%. These results are essential for artificial intelligence systems to understand and process information correctly. It shows that scientists and researchers can access information faster and accelerate scientific discovery using Turkish sources. Such artificial intelligence models can also be essential in education, providing students with a more effective and personalized learning experience. Therefore, processing Turkish chemistry and physics texts with artificial intelligence systems is essential in including studies conducted in this language in global studies in scientific research, education, and industrial applications.

Core genes and immune dysregulation in primary open-angle glaucoma: A molecular insight

Background Primary open-angle glaucoma (POAG) is a chronic, progressive and irreversible eye disease. Currently, there is no effective way to prevent optic nerve damage. Objective This study explored POAG gene markers to identify high-risk groups at an early stage and to find new effective therapeutic targets. Methods The mRNA and clinical information of POAG patients and normal samples were downloaded from the Gene Expression Omnibus (GEO) database. Through Weighted correlation network analysis (WGCNA) and generalized linear models (GLM), random forests (RF), support vector machines (SVM), and extreme gradient boosting (xGB) models, key risk genes were identified and an early diagnosis model was established. Functional enrichment analysis and CIBERSORT algorithm were used to further reveal the changes in the POAG immune environment and find emerging therapeutic targets. Results HERPUD1, IQCK, MRPL40, SRSF7 and TMEM243 were identified as risk genes, and the prediction model and nomogram constructed based on them had good early prediction efficiency. At the mechanistic level, the heterogeneity of T cell subsets seems to be a key factor affecting the progression of POAG and has potential therapeutic value. Conclusions HERPUD1, IQCK, MRPL40, SRSF7, and TMEM243 are of great significance for the early prediction and disease progression of POAG and have the potential value of becoming therapeutic targets.