Support Vector Machine Analysis Research Articles

A Comparative Analysis of Random Forest and Support Vector Machines for Classifying Irrigated Cropping Areas in The Upper-Comoé Basin, Burkina Faso

Objectives: This study investigates the performance of two machine-learning algorithms in classifying land areas across the Upper-Comoé basin in Burkina Faso. Methods: Within the Google Earth Engine data processing environment, Support Vector Machine (SVM) and the Random Forest (RF) algorithms were applied to a Landsat-8 OLI image of March 2019, to discriminate agricultural land areas, with an emphasis on irrigated areas. Findings: The results indicated good to excellent classification performance, with overall accuracies and Kappa coefficients between 71% and 99%, and 0.66 and 0.99, respectively. The RF method outperformed the SVM in terms of mapping "accuracy", but in terms of spatial distribution of classes, the SVM method provided a mapping close to reality, due to the density of the classes generated. Novelty: Our findings suggest that remote sensing can constitute a tool fully adapted to the needs of services in charge of agricultural water management in Burkina Faso. Keywords: Irrigation, Random Forest, Support Vector Machine, Google Earth Engine, Burkina Faso

Urogenital microbiota-driven virulence factor genes associated with recurrent urinary tract infection.

Urinary tract infections (UTIs) are a common health issue affecting individuals worldwide. Recurrent urinary tract infections (rUTI) pose a significant clinical challenge, with limited understanding of the underlying mechanisms. Recent research suggests that the urobiome, the microbial community residing in the urinary tract, may play a crucial role in the development and recurrence of urinary tract infections. However, the specific virulence factor genes (VFGs) driven by urobiome contributing to infection recurrence remain poorly understood. Our study aimed to investigate the relationship between urobiome driven VFGs and recurrent urinary tract infections. By analyzing the VFGs composition of the urinary microbiome in patients with rUTI compared to a control group, we found higher alpha diversity in rUTI patients compared with healthy control. And then, we sought to identify specific VFGs features associated with infection recurrence. Specifically, we observed an increased abundance of certain VGFs in the recurrent infection group. We also associated VFGs and clinical data. We then developed a diagnostic model based on the levels of these VFGs using random forest and support vector machine analysis to distinguish healthy control and rUIT, rUTI relapse and rUTI remission. The diagnostic accuracy of the model was assessed using receiver operating characteristic curve analysis, and the area under the ROC curve were 0.83 and 0.75. These findings provide valuable insights into the complex interplay between the VFGs of urobiome and recurrent urinary tract infections, highlighting potential targets for therapeutic interventions to prevent infection recurrence.

Identification and Analysis of Gene Biomarkers for Ovarian Cancer.

Objective: To identify potential diagnostic markers for ovarian cancer (OC) and explore the contribution of immune cells infiltration to the pathogenesis of OC. Methods: As the study cohort, two gene expression datasets of human OC (GSE27651 and GSE26712, taken as the metadata) taken from the Gene Expression Omnibus (GEO) database were combined, comprising 228 OC and 16 control samples. Analysis was performed to identify the differentially expressed genes between the OC and control samples, while support vector machine analysis using the recursive feature elimination algorithm and least absolute shrinkage and selection operator regression were performed to identify candidate biomarkers that could discriminate OC. In addition, immunohistochemistry staining was performed to verify the diagnostic value and protein expression levels of the candidate biomarkers. The GSE146553 dataset (OC n = 40, control n = 3) was used to further validate the diagnostic values of those biomarkers. Further, the proportions of various immune cells infiltration in the OC and control samples were evaluated using the CIBERSORT algorithm. Results: CLEC4M, PFKP, and SCRIB were identified as potential diagnostic markers for OC in both the metadata (area under the receiver operating characteristic curve [AUC] = 0.996, AUC = 1.000, AUC = 1.000) and GSE146553 dataset (AUC = 0.983, AUC = 0.975, AUC = 0.892). Regarding immune cell infiltration, there was an increase in the infiltration of follicular helper dendritic cells, and a decrease in the infiltration of M2 macrophages and neutrophils, as well as activated natural killer (NK) cells and T cells in OC. CLEC4M showed a significantly positive correlation with neutrophils (r = 0.57, p < 0.001) and resting NK cells (r = 0.42, p = 0.0047), but a negative correlation with activated dendritic cells (r = -0.33, p = 0.032). PFKP displayed a significantly positive correlation with activated NK cells (r = 0.36, p = 0.016) and follicular helper T cells (r = 0.32, p = 0.035), but a negative correlation with the naive B cells (r = -0.3, p = 0.049) and resting NK cells (r = -0.41, p = 0.007). SCRIB demonstrated a significantly positive correlation with plasma cells (r = 0.39, p = 0.01), memory B cells (r = 0.34, p = 0.025), and follicular helper T cells (r = 0.31, p = 0.04), but a negative correlation with neutrophils (r = -0.46, p = 0.002) and naive B cells (r = -0.48, p = 0.0012). Conclusion: CLEC4M, PFKP, and SCRIB were identified and verified as potential diagnostic biomarkers for OC. This work and identification of the three biomarkers may provide guidance for future studies into the mechanism and treatment of OC.

Fall Detection Using FMCW Radar to Reduce Detection Errors for the Elderly

Fall accidents pose a significant threat of severe injuries for the elderly, who often need immediate assistance when they fall. Since the use of conventional contact sensors or cameras might be uncomfortable for the user, research on fall detection using non-contact sensors has received considerable attention. While most prior studies have relied heavily on Doppler-based velocity parameters to detect falls, using only Doppler information may lead to erroneous detection of fall-like behavior. As a result, a feature that accounts for additional information is necessary. Addressing this need, this study developed an algorithm for classifying falls by detecting human motions using frequency modulation continuous wave radar, proposing a novel feature to reduce detection errors. The suggested feature was computed using the rangevelocity map of the 2D Fourier transform and evaluated using supervised machine learning techniques, such as support vector machine and linear discriminant analysis, attaining an accuracy higher than 91%.

Insight into the role of mitochondrion-related gene anchor signature in mitochondrial dysfunction of neutrophilic asthma

Objective At present, targeting molecular-pharmacological therapy is still difficult in neutrophilic asthma. The investigation aims to identify and validate mitochondrion-related gene signatures for diagnosis and specific targeting therapeutics in neutrophilic asthma. Methods Bronchial biopsy samples of neutrophilic asthma and healthy people were identified from the GSE143303 dataset and then matched with human mitochondrial gene data to obtain mitochondria-related differential genes (MitoDEGs). Signature mitochondria-related diagnostic markers were jointly screened by support vector machine (SVM) analysis, least absolute shrinkage, and selection operator (LASSO) regression. The expression of marker MitoDEGs was evaluated by validation datasets GSE147878 and GSE43696. The diagnostic value was evaluated by receiver operating characteristic (ROC) curve analysis. Meanwhile, the infiltrating immune cells were analyzed by the CIBERSORT. Finally, oxidative stress level and mitochondrial functional morphology for asthmatic mice and BEAS-2B cells were evaluated. The expression of signature MitoDEGs was verified by qPCR. Results 67 MitoDEGs were identified. Five signature MitoDEGs (SOD2, MTHFD2, PPTC7, NME6, and SLC25A18) were further screened out. The area under the curve (AUC) of signature MitoDEGs presented a good diagnostic performance (more than 0.9). There were significant differences in the expression of signature MitoDEGs between neutrophilic asthma and non-neutrophilic asthma. In addition, the basic features of mitochondrial dysfunction were demonstrated by in vitro and in vivo experiments. The expression of signature MitoDEGs in the neutrophilic asthma mice presented a significant difference from the control group. Conclusions These MitoDEGs signatures in neutrophilic asthma may hold potential as anchor diagnostic and therapeutic targets in neutrophilic asthma.

Reflectance spectra for identifying stress in different parts of leaf: a case study on oil palm seedlings

ABSTRACT In this study, the spectral responses to drought stress of different parts in the leaf of oil palm seedlings named base, middle, and tip were investigated. The ability to detect stress even before symptoms emerge requires knowledge of which part of oil palm leaves responds more quickly to the stress. The analysis of the reflectance spectra in region 650–1050 nm was conducted on respective sections of the leaves of the oil palm seedlings. The analysis revealed that the stress affects the tip of the leaf, followed by the middle and then the base. It was noticed that the greatest loss of water and chlorophyll content was at the tip of the leaf. Principal component analysis (PCA) and hierarchical cluster analysis (HCA) were used for clustering, while support vector machine (SVM) and linear discriminant analysis (LDA) were applied for categorization purposes. The outcomes of the PCA and HCA showed that the separation between the samples was based on the day and stress levels at respective sections of the leaf. With this, the possession of distinct morphological and physiological features by each part of the leaf can be concluded. From the PCA loadings, it was found that the regions 699–756 nm, 833–877 nm, and 933–958 nm showed noticeable bands and can be used to distinguish between the oil palm seedlings under stress. In addition, LDA and SVM classifiers demonstrated that the prediction accuracy of the stress level in oil palm seedlings was not influenced by the location in the leaf where the spectra were acquired.

Predicting academic achievement from the collaborative influences of executive function, physical fitness, and demographic factors among primary school students in China: ensemble learning methods

BackgroundElevated levels of executive function and physical fitness play a pivotal role in shaping future quality of life. However, few studies have examined the collaborative influences of physical and mental health on academic achievement. This study aims to investigate the key factors that collaboratively influence primary school students' academic achievement from executive function, physical fitness, and demographic factors. Additionally, ensemble learning methods are employed to predict academic achievement, and their predictive performance is compared with individual learners.MethodsA cluster sampling method was utilized to select 353 primary school students from Huai'an, China, who underwent assessments for executive function, physical fitness, and academic achievement. The recursive feature elimination cross-validation method was employed to identify key factors that collaboratively influence academic achievement. Ensemble learning models, utilizing eXtreme Gradient Boosting and Random Forest algorithms, were constructed based on Bagging and Boosting methods. Individual learners were developed using Support Vector Machine, Decision Tree, Logistic Regression, and Linear Discriminant Analysis algorithms, followed by the establishment of a Stacking ensemble learning model.ResultsOur findings revealed that sex, body mass index, muscle strength, cardiorespiratory function, inhibition, working memory, and shifting were key factors influencing the academic achievement of primary school students. Moreover, ensemble learning models demonstrated superior predictive performance compared to individual learners in predicting academic achievement among primary school students.ConclusionsOur results suggest that recognizing sex differences and emphasizing the simultaneous development of cognition and physical well-being can positively impact the academic development of primary school students. Ensemble learning methods warrant further attention, as they enable the establishment of an accurate academic early warning system for primary school students.

Development and validation of glycosyltransferase related-gene for the diagnosis and prognosis of head and neck squamous cell carcinoma.

Head and neck squamous cell carcinoma (HNSCC) is a highly heterogeneous cancer characterized by difficulties in early diagnosis and outcome prediction. Aberrant glycosylated structures produced by the aberrant expression of glycosyltransferases are prevalent in HNSCC. In this study, we aim to construct glycosyltransferase-related gene signatures with diagnostic and prognostic value to better stratify patients with HNSCC and improve their diagnosis and prognosis. Bioinformatic tools were used to process data of patients with HNSCC from The Cancer Genome Atlas (TCGA) database. The prognostic model was formatted using univariate and multivariate Cox regression methods, while the diagnostic signature was constructed using support vector machine (SVM) and LASSO analysis. The results were verified using the Gene Expression Omnibus (GEO) cohort. The tumor microenvironment and benefits of immune checkpoint inhibitor (ICI) therapy in subgroups defined by glycosyltransferase-related genes were analyzed. Molecular biology experiments, including western blotting, cell counting kit (CCK)-8, colony formation, wound healing, and Transwell assays, were conducted to confirm the oncogenic function of beta-1,4-galactosyltransferase 3 (B4GALT3) in HNSCC. We established a five-gene prognostic signature and a 15-gene diagnostic model. Based on the median risk score, patients with low risk had longer overall survival than those in the high-risk group, which was consistent with the results of the GEO cohort. The concrete results suggested that high-risk samples were related to a high tumor protein (TP)53 mutation rate, high infiltration of resting memory cluster of differentiation (CD)4 T cells, resting natural killer (NK) cells, and M0 macrophages, and benefited from ICI therapy. In contrast, the low-risk subgroup was associated with a low TP53 mutation rate; and high infiltration of naive B cells, plasma cells, CD8 T cells, and resting mast cells; and benefited less from ICI therapy. In addition, the diagnostic model had an area under curve (AUC) value of 0.997 and 0.978 in the training dataset and validation cohort, respectively, indicating the high diagnostic potential of the model. Ultimately, the depletion of B4GALT3 significantly hindered the proliferation, migration, and invasion of HNSCC cells. We established two new biomarkers that could provide clinicians with diagnostic, prognostic, and treatment guidance for patients with HNSCC.

Central autonomic network and early prognosis in patients with disorders of consciousness

The central autonomic network (CAN) plays a crucial role in modulating the autonomic nervous system. Heart rate variability (HRV) is a valuable marker for assessing CAN function in disorders of consciousness (DOC) patients. We used HRV analysis for early prognosis in 58 DOC patients enrolled within ten days of hospitalization. They underwent a five-minute electrocardiogram during baseline and acoustic/visual stimulation. The coma recovery scale-revised (CRS-R) was used to define the patient’s consciousness level and categorize the good/bad outcome at three months. The high-frequency Power Spectrum Density and the standard deviation of normal-to-normal peaks in baseline, the sample entropy during the stimulation, and the time from injury features were used in the support vector machine analysis (SVM) for outcome prediction. The SVM predicted the patients’ outcome with an accuracy of 96% in the training test and 100% in the validation test, underscoring its potential to provide crucial clinical information about prognosis.

Parrot Species Recognition by using CNN &amp; SVM

This research project addresses the imperative task of parrot species recognition through the integration of Convolutional Neural Network (CNN) and Support Vector Machine (SVM) methodologies. Focused on three distinct parrot species—'Blue and Gold Macaw,' 'Ringneck Parakeet,' and 'Sun Conure'—the study explores the efficacy of advanced machine learning techniques in ornithological research. The CNN serves as a robust feature extractor, autonomously learning intricate visual patterns crucial for species identification. Meanwhile, the SVM acts as a discriminative classifier, optimizing decision boundaries in the feature space. Comparative analysis reveals the CNN's superior accuracy of 95%, surpassing the SVM's 83.33%. The visual representation, a bar chart, provides a clear and accessible depiction of the research findings. Titled "Comparative Analysis of CNN and SVM for Accuracy Assessment of Parrot Species," the chart ensures relevance and context for the audience. Overall, this research contributes valuable insights into the potential of deep learning in parrot species recognition, with implications for automated wildlife monitoring and biodiversity conservation in the realm of ornithology.

Age prediction using DNA methylation of Y-chromosomal CpGs in semen samples

In cases of sexual assault, the evidence often exists as a mixture of female and male body fluids, and in many cases, contains a higher proportion of female body fluids than males. In these cases, Y-STR, rather than autosomal STRs, can provide useful information. It becomes very difficult to identify the true suspect if there is no match among known suspects or if a match exists for two or more suspects, e.g. two suspects from the same paternal lineage. However, age prediction using the DNA methylation of Y-chromosomal CpGs can help narrow the search for unknown suspects and discriminate between older and younger suspects. Therefore, the DNA methylation profiles of semen samples from 56 healthy Korean males were generated using Illumina’s Infinium MethylationEPIC BeadChip Array. Among the ten identified age-associated CpG markers located in the Y-chromosome, nine were used to construct age prediction models. The identified markers were further investigated in the MPS analysis of 147 semen samples, and the multiplex assay was validated with the reliability, reproducibility and sensitivity tests. Several age prediction models were constructed using the MPS data with the multiple linear regression, stepwise linear regression, ridge linear regression, lasso regression, elastic net linear regression and support vector machine analyses, and all showed MAEs of 5 to 7 years in the test set samples. Six single-source female samples were also subjected to MPS analysis but showed very low coverage that could not affect the analysis of the mixed samples. Therefore, the age prediction models of the present study are expected to provide useful investigative leads, especially in mixed male and female samples from sexual assault cases.

A Comparative Analysis of Support Vector Machine and Decision Tree Algorithm for Predicting Fault in Uninterruptible Power Supply Systems

A Comparative Analysis of Support Vector Machine and Decision Tree Algorithm for Predicting Fault in Uninterruptible Power Supply Systems

Performance Analysis of Support Vector Machine (SVM) on Challenging Datasets for Forest Fire Detection

Performance Analysis of Support Vector Machine (SVM) on Challenging Datasets for Forest Fire Detection

Integrated Bioinformatics Analysis Reveals the Aberrantly Methylated Differentially Expressed Genes in Dilated Cardiomyopathy.

Dilated cardiomyopathy (DCM) causes heart failure and sudden death. Epigenetics is crucial in cardiomyopathy susceptibility and progression; however, the relationship between epigenetics, particularly DNA methylation, and DCM remains unknown. Therefore, this study identified aberrantly methylated differentially expressed genes (DEGs) associated with DCM using bioinformatics analysis and characterized their clinical utility in DCM. DNA methylation expression profiles and transcriptome data from public datasets of human DCM and healthy control cardiac tissues were obtained from the Gene Expression Omnibus public datasets. Then an epigenome-wide association study was performed. DEGs were identified in both DCM and healthy control cardiac tissues. In total, 3,353 cytosine-guanine dinucleotide sites annotated to 2,818 mRNAs were identified, and 479 DCM-related genes were identified. Subsequently, core genes were screened using logistic, least absolute shrinkage and selection operator, random forest, and support vector machine analyses. The overlapping of these genes resulted in DEGs with abnormal methylation patterns. Cross-tabulation analysis identified 8 DEGs with abnormal methylation. Real-time quantitative polymerase chain reaction confirmed the expression of aberrantly methylated DEGs in mice. In DCM murine cardiac tissues, the expressions of SLC16A9, SNCA, PDE5A, FNDC1, and HTRA1 were higher compared to normal murine cardiac tissues. Moreover, logistic regression model associated with aberrantly methylated DEGs was developed to evaluate the diagnostic value, and the area under the receiver operating characteristic curve was 0.949, indicating that the diagnostic model could reliably distinguish DCM from non-DCM samples. In summary, our study identified 5 DEGs through integrated bioinformatic analysis and in vivo experiments, which could serve as potential targets for further comprehensive investigation.

Comparative analysis of k-nearest neighbor and support vector machine in classification of COVID 19 disease in Makassar City

Comparative analysis of k-nearest neighbor and support vector machine in classification of COVID 19 disease in Makassar City

Multi-ensemble machine learning framework for omics data integration: A case study using breast cancer samples

Integration of voluminous omics data aids to unravel biological complexities associated with different disease phenotypes. Machine learning (ML) approaches provide insightful techniques for systematic multi-omics data integration. In this study, survival prediction of breast cancer patients was undertaken using omics data of 302 female patients from The Cancer Genome Atlas (TCGA). The data included gene expression, miRNA expression, DNA methylation and copy number variation. Three computational multi-ensemble ML pipelines were tested using Support Vector Machine (SVM), Random Forest (RF) and Partial Least Squares-Discriminant Analysis (PLS-DA) algorithms. To overcome the limitations associated with univariate feature selection criteria, the ML pipelines were built along with latent factors obtained by multivariate dimension reduction method. This facilitated investigation of background genetic networks and identification of potential hub genes. Analysis of the results obtained revealed that SVM with PLS-DA method (integrated with gene expression, DNA methylation, and miRNA expression modalities) was the best-performing model with an Area Under Curve (AUC) of 89% and an accuracy of 83% for survival prediction. This study not only corroborated previously reported breast cancer-specific prognostic biomarkers but also predicted additional potential biomarkers. The work demonstrates the effective use of a multi-ensemble ML model with efficient feature selection methods as a robust protocol for cancer genotype to phenotype correlation.

A Survey on Scenario Theory, Complexity, and Compression-Based Learning and Generalization.

This work investigates formal generalization error bounds that apply to support vector machines (SVMs) in realizable and agnostic learning problems. We focus on recently observed parallels between probably approximately correct (PAC)-learning bounds, such as compression and complexity-based bounds, and novel error guarantees derived within scenario theory. Scenario theory provides nonasymptotic and distributional-free error bounds for models trained by solving data-driven decision-making problems. Relevant theorems and assumptions are reviewed and discussed. We propose a numerical comparison of the tightness and effectiveness of theoretical error bounds for support vector classifiers trained on several randomized experiments from 13 real-life problems. This analysis allows for a fair comparison of different approaches from both conceptual and experimental standpoints. Based on the numerical results, we argue that the error guarantees derived from scenario theory are often tighter for realizable problems and always yield informative results, i.e., probability bounds tighter than a vacuous [0,1] interval. This work promotes scenario theory as an alternative tool for model selection, structural-risk minimization, and generalization error analysis of SVMs. In this way, we hope to bring the communities of scenario and statistical learning theory closer, so that they can benefit from each other's insights.

Predicting Turbidite Channel in Deep-Water Canyon Based on Grey Relational Analysis-Support Vector Machine Model: A Case Study of the Lingshui Depression in Qiongdongnan Basin, South China Sea

Predicting Turbidite Channel in Deep-Water Canyon Based on Grey Relational Analysis-Support Vector Machine Model: A Case Study of the Lingshui Depression in Qiongdongnan Basin, South China Sea

Abnormal noise identification of engines based on wavelet packet transform and bispectrum analysis

Abnormal noise is the most prominent problem for motorcycles and affects the consumers’ purchasing desire and driving experience and the enterprise’s competitiveness. Usually, the noise from a newly assembled engine is detected by manual auscultation (MA) to determine if the engine is operating normally. However, MA is also affected by subjective and objective factors with severe labor intensity, and its accuracy greatly fluctuates. Importantly, MA cannot be applied in a corporation with mass production and high-quality requirements. To improve the efficiency and accuracy of motorcycle engine quality inspection and achieve intelligent production, an online engine abnormal noise detection method was proposed based on wavelet packet transform (WPT) and bispectrum analysis (BA); this method improved the accuracy and stability of the identification of the abnormal noise engine and reduced the cost of the check. First, the acoustic signal of the engine of the motorcycle was acquired by using a free-field microphone. Second, the background noise of signals was eliminated by using the wavelet correlation coefficient (WCC) theory, and the signal features were extracted by applying WPT and BA. Third, the feature vectors were normalized before being used as support vector machine (SVM) samples. Fourth, an appropriate kernel function and parameters were selected to train the vector machine using the training sets. Finally, the testing sets were used to inspect the accuracy of the vector machines. The result showed that the training accuracy is 95% and the testing accuracy is 97.5 of the samples were suitable by using the method of wavelet packet transform-bispectrum analysis-support vector machines (WPT-BA-SVM). WPT-BA-SVM effectively identified engine fault types and provided the theoretical foundation for the establishment of an engine abnormal noise online detection system.

Feature Extraction Evaluation of Various Machine Learning Methods for Finger Movement Classification using Double Myo Armband

The deployment of electromyography (EMG) signals attracts many researchers since it can be used in decoding finger movements for exoskeleton robotics, prosthetics hand, and powered wheelchair. However, decoding any movement is a challenging task. The success of EMG signals' use lies in the appropriate choice of feature extraction and classification model, especially in the feature extraction process. Therefore, this study evaluates an eight-feature extraction evaluation on various machine learnings such as the Support Vector Machine (SVM), k-Nearest Neighbor (k-NN), Decision Tree (DT), Naïve Bayes (NB), and Quadratic Discriminant Analysis (QDA). The dataset from four intact subjects is used to classify twelve finger movements. Through 5 cross-validations, the result shows that almost all feature extractions combined with SVM outperform other combinations of features and classifiers. Mean Absolute Value (MAV) as a feature and SVM as a classifier highlight the best combination with an accuracy of 94.01%.