A shadow detection and removal method for fruit recognition in natural environments

In a DNA microarray dataset, gene expression data often has a huge number of features(which are referred to as genes) versus a small size of samples. With the development of DNA microarray technology, the number of dimensions increases even faster than before, which could lead to the problem of the curse of dimensionality. To get good classification performance, it is necessary to preprocess the gene expression data. Support vector machine recursive feature elimination (SVM-RFE) is a classical method for gene selection. However, SVM-RFE suffers from high computational complexity. To remedy it, this paper enhances SVM-RFE for gene selection by incorporating feature clustering, called feature clustering SVM-RFE (FCSVM-RFE). The proposed method first performs gene selection roughly and then ranks the selected genes. First, a clustering algorithm is used to cluster genes into gene groups, in each which genes have similar expression profile. Then, a representative gene is found to represent a gene group. By doing so, we can obtain a representative gene set. Then, SVM-RFE is applied to rank these representative genes. FCSVM-RFE can reduce the computational complexity and the redundancy among genes. Experiments on seven public gene expression datasets show that FCSVM-RFE can achieve a better classification performance and lower computational complexity when compared with the state-the-art-of methods, such as SVM-RFE.

Metabolic markers are the core of metabonomic surveys. Hence selection of differential metabolites is of great importance for either biological or clinical purpose. Here, a feature selection method was developed for complex metabonomic data set. As an effective tool for metabonomics data analysis, support vector machine (SVM) was employed as the basic classifier. To find out meaningful features effectively, support vector machine recursive feature elimination (SVM-RFE) was firstly applied. Then, genetic algorithm (GA) and random forest (RF) which consider the interaction among the metabolites and independent performance of each metabolite in all samples, respectively, were used to obtain more informative metabolic difference and avoid the risk of false positive. A data set from plasma metabonomics study of rat liver diseases developed from hepatitis, cirrhosis to hepatocellular carcinoma was applied for the validation of the method. Besides the good classification results for 3 kinds of liver diseases, 31 important metabolites including lysophosphatidylethanolamine (LPE) C16:0, palmitoylcarnitine, lysophosphatidylethanolamine (LPC) C18:0 were also selected for further studies. A better complementary effect of the three feature selection methods could be seen from the current results. The combinational method also represented more differential metabolites and provided more metabolic information for a “global” understanding of diseases than any single method. Further more, this method is also suitable for other complex biological data sets.

Feature selection and analysis on correlated gas sensor data with recursive feature elimination

Effects of imaging modalities, brain atlases and feature selection on prediction of Alzheimer's disease

BackgroundEndometriosis is recognized as an estrogen-dependent inflammation disorder, estimated to affect 8%-15% of women of childbearing age. Currently, the etiology and pathogenesis of endometriosis are not completely clear. Underlying mechanism for endometriosis is still under debate and needs further exploration. The involvement of transcription factors and immune mediations may be involved in the pathophysiological process of endometriosis, but the specific mechanism remains to be explored. This study aims to investigate the underlying molecular mechanisms in endometriosis.MethodsThe gene expression profile of endometriosis was obtained from the gene expression omnibus (GEO) database. Gene set variation analysis (GSVA) and gene set enrichment analysis (GSEA) were applied to the endometriosis GSE7305 datasets. Cibersort and MCP-counter were used to explore the immune response gene sets, immune response pathway, and immune environment. Differentially expressed genes (DEGs) were identified and screened. Common biological pathways were being investigated using the kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis. Transcription factors were from The Human Transcription Factors. The least absolute shrinkage and selection operator (Lasso) model identified four differential expressions of transcription factors (AEBP1, HOXB6, KLF2, and RORB). Their diagnostic value was calculated by receiver operating characteristic (ROC) curve analysis and validated in the validation cohort (GSE11691, GSE23339). By constructing the interaction network of crucial transcription factors, weighted gene coexpression network analysis (WGCNA) was used to search for key module genes. Metascape was used for enrichment analysis of essential module genes and obtained HOXB6, KLF2. The HOXB6 and KLF2 were further verified as the only two intersection genes according to Support Vector Machine Recursive Feature Elimination (SVM-RFE) and random forest models. We constructed ceRNA (lncRNA-miRNA-mRNA) networks with four potential transcription factors. Finally, we performed molecular docking for goserelin and dienogest with four transcription factors (AEBP1, HOXB6, KLF2, and RORB) to screen potential drug targets.ResultsImmune and metabolic pathways were enriched in GSVA and GSEA. In single sample gene set enrichment analysis (ssGSEA), most immune infiltrating cells, immune response gene sets, and immune response pathways are differentially expressed between endometriosis and non-endometriosis. Twenty-seven transcription factors were screened from differentially expressed genes. Most of the twenty-seven transcription factors were correlated with immune infiltrating cells, immune response gene sets and immune response pathways. Furthermore, Adipocyte enhancer binding protein 1 (AEBP1), Homeobox B6 (HOXB6), Kruppel Like Factor 2 (KLF2) and RAR Related Orphan Receptor B (RORB) were selected out from twenty-seven transcription factors. ROC analysis showed that the four genes had a high diagnostic value for endometriosis. In addition, KLF2 and HOXB6 were found to play particularly important roles in multiple modules (String, WGCNA, SVM-RFE, random forest) on the gene interaction network. Using the ceRNA network, we found that NEAT1 may regulate the expressions of AEBP1, HOXB6 and RORB, while X Inactive Specific Transcript (XIST) may control the expressions of HOXB6, RORB and KLF2. Finally, we found that goserelin and dienogest may be potential drugs to regulate AEBP1, HOXB6, KLF2 and RORB through molecular docking.ConclusionsAEBP1, HOXB6, KLF2, and RORB may be potential biomarkers for endometriosis. Two of them, KLF2 and HOXB6, are critical molecules in the gene interaction network of endometriosis. Discovered by molecular docking, AEBP1, HOXB6, KLF2, and RORB are targets for goserelin and dienogest.

Skin cancer is on the rise. Hence the accurate detection of cancerous lesions is of paramount importance in the treatment of this health condition. In this study, we present a computer vision framework that studies 10 skin lesion features including newly introduced morphological and texture features along with established in the literature. All features are extracted automatically from 90 lesion images. A two-class classification problem was applied to determine the most significant features of disease using three features selection methods, Support Vector Machines Recursive Feature Elimination (SVMRFE), Information Gain, and Correlation-based Feature Subset Selection. We found that the five features selected by SVMRFE provides the highest accuracy readings of 100% model, 84% leave-one-out, and 89% 10-fold cross validation (10×CV) than the other two methods. Comparing the selected features with those used by the Total Dermoscopy Score, we report consistencies, disagreements, and the contributions of the present work.

Selecting distinguishable features from small sample data of rolling bearings has become an emerging and trending research topic in recent years. In this paper, a two-stage feature selection method based on Euclidean distance and multi-class support vector machine is proposed to remove irrelevant features and redundant features. First, the fault data are formed to a multidomain feature set. Second, the ReliefF algorithm based on Euclidean distance is used to remove irrelevant features. Third, the support vector machine recursive feature elimination (SVMRFE) is used to remove redundant features. Lastly, the KNN classifier is used to verify the effectiveness of this method. This paper uses the rolling bearing public datasets. After comparison, the results prove that the method proposed in the article is better.

IntroductionIn the classification of Mass Spectrometry (MS) proteomics data, peak detection, feature selection, and learning classifiers are critical to classification accuracy. To better understand which methods are more accurate when classifying data, some publicly available peak detection algorithms for Matrix assisted Laser Desorption Ionization Mass Spectrometry (MALDI-MS) data were recently compared; however, the issue of different feature selection methods and different classification models as they relate to classification performance has not been addressed. With the application of intelligent computing, much progress has been made in the development of feature selection methods and learning classifiers for the analysis of high-throughput biological data. The main objective of this paper is to compare the methods of feature selection and different learning classifiers when applied to MALDI-MS data and to provide a subsequent reference for the analysis of MS proteomics data.ResultsWe compared a well-known method of feature selection, Support Vector Machine Recursive Feature Elimination (SVMRFE), and a recently developed method, Gradient based Leave-one-out Gene Selection (GLGS) that effectively performs microarray data analysis. We also compared several learning classifiers including K-Nearest Neighbor Classifier (KNNC), Naïve Bayes Classifier (NBC), Nearest Mean Scaled Classifier (NMSC), uncorrelated normal based quadratic Bayes Classifier recorded as UDC, Support Vector Machines, and a distance metric learning for Large Margin Nearest Neighbor classifier (LMNN) based on Mahanalobis distance. To compare, we conducted a comprehensive experimental study using three types of MALDI-MS data.ConclusionRegarding feature selection, SVMRFE outperformed GLGS in classification. As for the learning classifiers, when classification models derived from the best training were compared, SVMs performed the best with respect to the expected testing accuracy. However, the distance metric learning LMNN outperformed SVMs and other classifiers on evaluating the best testing. In such cases, the optimum classification model based on LMNN is worth investigating for future study.

As an intermediary between residential customers and system operators, load aggregators (LA) are responsible for integrating the residential customers' demand response (DR) potential and enabling their transactions in the day-ahead market. Profit of the market trading process mainly relies on the pre-estimated achievable DR potential, which puts forwards the necessity of its accurate forecasting. Therefore, this paper proposes a probabilistic forecasting model to forecast the day-ahead achievable DR potential at the aggregated level under an incentive-based demand response (IBDR) program. Firstly, we attempt to establish the customers' DR potential, during which a home energy management system (HEMS) is introduced to implement load adjustment for electrical appliances. Secondly, several features that may affect the DR potential are extracted, among which the more relevant ones are selected through the support vector machine recursive feature elimination (SVM-RFE) method. Finally, based on these selected features, a support vector machine (SVM) method is adopted to establish the DR potential point forecasting model, and then the probabilistic forecasting model of the aggregated DR potential is established through the superimposition of the point forecasting results and the corresponding error distribution, the latter could be estimated by the non-parametric kernel density (NKDE) method. Case studies show that a good performance could be achieved by the proposed probabilistic forecasting model and the feature selection process could significantly improve the forecasting accuracy.

With the massive expansion of information in social media, a high demand exists for information retrieval techniques. The recognition of the named entity (NE) such as person, location, organization, etc. has emerged as one of the main tasks in natural language processing. Often, utilizing the entire feature set may not only be time-consuming but may also have a negative effect on performance. Due to the high number of features, it is difficult to identify the subset of features relevant to a given task. In this paper, we apply feature selection methods based on the support vector machine recursive feature elimination (SVM-RFE) to find the optimized feature set. Afterward, an optimized feature set combination is used to identify and classify named entities (NEs) based on the Support Vector Machine (SVM). The proposed method is evaluated using Darwish’s dataset (a publicly available benchmark for Arabic NER for social media). Experimental results demonstrate the effectiveness of feature selection in enhancing performance and outperform the most advanced systems

Osteoporosis (OP) is a prevalent metabolic disease, with aging and menopause being the major risk factors. Studies have shown that nearly one-third of postmenopausal women suffer from osteoporosis. However, there is a scarcity of research on antioxidant systems for the prevention and treatment of postmenopausal osteoporosis (PM-OP). To address this gap, we performed differential analysis using Limma to identify differentially expressed genes (DEGs) in PM-OP samples. We employed weighted correlation network analysis (WGCNA) to identify oxidative stress (OS)-related genes (OSRGs) highly correlated with PM-OP. The intersection of key modular genes and DEGs yielded differentially expressed OSRGs (DE-OSRGs) specific to PM-OP. We conducted GO and KEGG functional enrichment analyses on these genes. Additionally, we constructed a protein-protein interaction (PPI) network and utilized support vector machine recursive feature elimination (SVM-RFE) and random forest (RF) algorithms to identify signature genes. The diagnostic value of the signature genes was evaluated and validated using ROC curves. GSEA enrichment analysis was employed to explore the potential mechanisms associated with the signature genes. Finally, we constructed a regulatory network involving TF-miRNA-mRNA interactions for the signature genes and verified the biological roles of FOXO3 and DDIT3 in PM-OP and healthy groups using quantitative real-time PCR (qRT-PCR). Our analysis revealed 20 DE-OSRGs specific to PM-OP, obtained by intersecting modular and differential genes. The PPI network identified central genes (DDIT3, MAPK8, CDK2, SIRT1, and FOXO3) with more than 3 nodes. Through integration with machine learning algorithms, we identified DDIT3 and FOXO3 as signature genes. The ROC curve analysis indicated that the AUC value was greater than 0.7, suggesting the potential diagnostic value of these signature genes. Furthermore, GSEA results revealed their involvement in pathways related to the regulation of neutrophil activation, oxidative phosphorylation, MAPK signaling, mitochondrial matrix, and phagocytosis. Lastly, we constructed a regulatory network comprising 27 nodes (22 TFs, 3 miRNAs, and 2 mRNAs) and 28 edges. Additionally, qRT-PCR confirmed the significant up-regulation of FOXO3 and DDIT3 expressions in the PM-OP group compared to the healthy control group. In summary, this study employed bioinformatics analysis to identify OS-related biomarkers (DDIT3 and FOXO3) in PM-OP, providing new biological targets for clinical treatment.

Arthritis, a common chronic disease encompassing multiple subtypes of osteoarthritis and rheumatoid arthritis, was explored in this study as a risk-related factor based on data from the 2023 U.S. National Health Interview Survey (NHIS). The study included 26,031 participants (6,849 in the arthritis group; 19,182 in the control group), and 21 variables were found to be significantly different between groups by chi-square test. Fourteen key predictors were screened using support vector machine recursive feature elimination (SVM-RFE): age, general health, chronic obstructive pulmonary disease, gender, hypertension, coronary heart disease, body mass index (BMI), cancer, depression, dementia, asthma, diabetes, smoking status, and hepatitis. The column-linear graphical model constructed based on these variables showed excellent predictive performance (AUC = 0.813), the slope of the calibration curve was close to 1 (P = 0.444) indicating high predictive accuracy, and the decision curve analysis showed that its net benefit was better than that of a single predictor. The study demonstrated that the NHIS column-line graph model constructed based on machine learning algorithms can effectively predict the risk of arthritis and provide an important reference for clinical management. The prediction model established in this study provides a theoretical basis for accurate prevention and treatment strategies for arthritis.

The issue of multi-drug-resistant tuberculosis (MDR-TB) presents a substantial challenge to global public health. Regrettably, the diagnosis of drug-resistant tuberculosis (DR-TB) frequently necessitates an extended period or more extensive laboratory resources. The swift identification of MDR-TB poses a particularly challenging endeavor. To identify the biomarkers indicative of multi-drug resistance, we conducted a screening of the GSE147689 dataset for differentially expressed genes (DEGs) and subsequently conducted a gene enrichment analysis. Our analysis identified a total of 117 DEGs, concentrated in pathways related to the immune response. Three machine learning methods, namely random forest, decision tree, and support vector machine recursive feature elimination (SVM-RFE), were implemented to identify the top 10 genes according to their feature importance scores. A4GALT and S1PR1, which were identified as common genes among the three methods, were selected as potential molecular markers for distinguishing between MDR-TB and drug-susceptible tuberculosis (DS-TB). These markers were subsequently validated using the GSE147690 dataset. The findings suggested that A4GALT exhibited area under the curve (AUC) values of 0.8571 and 0.7121 in the training and test datasets, respectively, for distinguishing between MDR-TB and DS-TB. S1PR1 demonstrated AUC values of 0.8163 and 0.5404 in the training and test datasets, respectively. When A4GALT and S1PR1 were combined, the AUC values in the training and test datasets were 0.881 and 0.7551, respectively. The relationship between hub genes and 28 immune cells infiltrating MDR-TB was investigated using single sample gene enrichment analysis (ssGSEA). The findings indicated that MDR-TB samples exhibited a higher proportion of type 1T helper cells and a lower proportion of activated dendritic cells in contrast to DS-TB samples. A negative correlation was observed between A4GALT and type 1T helper cells, whereas a positive correlation was found with activated dendritic cells. S1PR1 exhibited a positive correlation with type 1T helper cells and a negative correlation with activated dendritic cells. Furthermore, our study utilized connectivity map analysis to identify nine potential medications, including verapamil, for treating MDR-TB. In conclusion, our research identified two molecular indicators for the differentiation between MDR-TB and DS-TB and identified a total of nine potential medications for MDR-TB.

Cyanobacteria have developed acclimation strategies to adapt to harsh environments, making them a model organism. Understanding the molecular mechanisms of tolerance to abiotic stresses can help elucidate how cells change their gene expression patterns in response to stress. Recent advances in sequencing techniques and bioinformatics analysis methods have led to the discovery of many genes involved in stress response in organisms. The Synechocystis sp. PCC 6803 is a suitable microorganism for studying transcriptome response under environmental stress. Therefore, for the first time, we employed two effective feature selection techniques namely and support vector machine recursive feature elimination (SVM-RFE) and LASSO (Least Absolute Shrinkage Selector Operator) to pinpoint the crucial genes responsive to environmental stresses in Synechocystis sp. PCC 6803. We applied these algorithms of machine learning to analyze the transcriptomic data of Synechocystis sp. PCC 6803 under distinct conditions, encompassing light, salt and iron stress conditions. Seven candidate genes namely sll1862, slr0650, sll0760, slr0091, ssl3044, slr1285, and slr1687 were selected by both LASSO and SVM-RFE algorithms. RNA-seq analysis was performed to validate the efficiency of our feature selection approach in selecting the most important genes. The RNA-seq analysis revealed significantly high expression for five genes namely sll1862, slr1687, ssl3044, slr1285, and slr0650 under ion stress condition. Among these five genes, ssl3044 and slr0650 could be introduced as new potential candidate genes for further confirmatory genetic studies, to determine their roles in their response to abiotic stresses.

Two-centers machine learning analysis for predicting acid-fast bacilli results in tuberculosis sputum tests.