Support Vector Machine Research Articles

Analysis of the Style Characteristics of Regional Folk Songs and Music Classification Algorithms

Regional folk songs have a rich history and are filled with cultural values. In this paper, first, the style characteristics of regional folk songs are briefly introduced. Using four regional folk songs from the northwest, northeast, southwest, and Hakka as examples, time domain, frequency domain, and mel-frequency cepstral coefficient (MFCC) features were extracted. Finally, the bidirectional long short-term memory (BiLSTM)-based music classification algorithm is used to realize the classification of folk songs from different regions. It was found that using the time-frequency domain + MFCC as features produced better results in music classification than using only the time-frequency domain or only MFCC features. The BiLSTM algorithm achieved an accuracy of 0.8339 and an F1 value of 0.8201 for the 10 s fragment set, both of which were better than those of the K-nearest neighbor, support vector machine, and other classification algorithms. The results show that the approach used in this study to categorize regional folk songs is reliable and that it can be applied to real folk songs.

Improved Back-Propagation Neural Network Applied to Enterprise Employee Performance Appraisal and Evaluation

Evaluating employee performance in enterprises is beneficial for improving talent competitiveness, operation level production, and overall enterprise performance. At present, many performance appraisal methods lack objectivity and fairness, which is not conducive to the long-term development of enterprises. Therefore, more research into scientific and effective performance appraisal methods is required. This study takes the grassroots employees of Enterprise A as an example, improving the existing performance appraisal evaluation indices to evaluate employee performance from three dimensions, including achievement, ability, and attitude, as determined by index weights using the analytic hierarchy process approach. An improved back-propagation neural network (BPNN) method is then designed to obtain performance appraisal and evaluation results. The error between the output of the improved BPNN method and expected output was small. Of the 20 extracted samples, the maximum and minimum error values were 0.05 and 0.01, respectively, and the average error was 0.03. The improved BPNN method evaluated only one out of 20 samples incorrectly, and the accuracy of the improved BPNN method was 96.21%, which is 19.88% and 10.75% higher than those of support vector machine and standard BPNN, respectively. The findings demonstrate that the improved BPNN method can be used in the appraisal and assessment of enterprise employee performance and has practical application value.

Recent Advancements in Anomaly Detection for Manufacturing Systems

Recent advancements in anomaly detection and root cause analysis have revolutionized manufacturing processes, enabling improved efficiency, quality control, and cost reduction. This review explores the latest developments in these fields, focusing on supervised and unsupervised anomaly detection methods, statistical techniques for root cause analysis, and predictive modeling approaches. Supervised anomaly detection methods, such as classification-based approaches, support vector machines, and decision trees, leverage labeled data to identify anomalies. Unsupervised techniques, including clustering-based methods (e.g., k-means and DBSCAN), density-based approaches (e.g., local outlier factor), and dimensionality reduction (e.g., PCA and autoencoders), detect anomalies without prior knowledge of anomalous instances. Statistical methods for root cause analysis, such as statistical process control, hypothesis testing, correlation analysis, and multivariate techniques (e.g., factor analysis), enable the identification of the underlying causes of manufacturing issues. Predictive modeling techniques, including time-series forecasting, machine learning for predictive maintenance, and optimization methods for process improvement, further enhance manufacturing efficiency. The integration of these techniques has led to significant improvements in quality control, equipment failure prediction, and supply chain optimization. However, challenges remain in handling complex, high-dimensional data and integrating domain knowledge with data-driven approaches. Future research directions include the development of explainable AI, edge computing, and federated learning for anomaly detection in manufacturing. This analysis offers a thorough examination of cutting-edge techniques for detecting anomalies and analyzing root causes, emphasizing their capacity to transform manufacturing operations and stimulate future advancements in these research areas. Keywords—anomaly detection, manufacturing, supervised and unsupervised anomaly detection

Identification of the rank and kind of ten coals including bituminous and anthracite based on near-infrared spectroscopy and interpretable GASF-ResNet18 method

ABSTRACT Different ranks and kinds of coal have different properties (volatile, ash, etc.), which affect the process parameters in clean utilization of coal. Based on near-infrared (NIR) spectroscopy data, the 1D spectral data of 670 coal samples were encoded into 2D images with Gramian Angular Summation Field (GASF) method as input data, and a convolutional neural network (CNN) known as ResNet18 was applied to identify the rank and kind of coal. On a dataset containing seven kinds of bituminous coal and three kinds of anthracite coal, the proposed GASF-ResNet18 method achieves an accuracy of 96.3% for coal rank identification, and 94.8% for coal kind identification, even in the presence of moisture interference. Compared with traditional machine learning methods (partial least squares discriminant analysis, back propagation network, logistic regression, and support vector machine), GASF-ResNet18 method has a higher accuracy. It was attributed to ResNet18 catching the key information provided by the following GASF image regions: the combination of the absorption rate in 1250–1700 nm in column and 1550–1700 nm in row, and the combination of 900–1150 nm in column and 1150–1450 nm in row. The results showed GASF-ResNet18 method based on NIR data can be used to rapidly identify the rank and kind of coal.

Early Detection of Alzheimer's Disease from MR Images Using Fine-Tuning Neighborhood Component Analysis and Convolutional Neural Networks

Abstract This study develops an automatic algorithm for detecting Alzheimer's disease (AD) using magnetic resonance imaging (MRI) through deep learning and feature selection techniques. It utilizes a dataset of 6400 MRI images from Kaggle, categorized into four classes. Initially, the study employs pretrained CNN architectures—DenseNet-201, MobileNet-v2, ResNet-18, ResNet-50, ResNet-101, and ShuffleNet—for classification using five fold cross-validation, with DenseNet-201 achieving the highest accuracy of 82.11%. Due to the dataset's size and imbalance, as well as the long training times, the study aims to create a more efficient algorithm. The CNNs are used as deep feature extractors from AD images, and the extracted features are reduced using a new fine-tuning neighborhood component analysis (FTNCA) algorithm, which minimizes loss and determines the optimal tolerance value. The essential features are then classified using various machine learning algorithms, including artificial neural network (ANN), K-nearest neighbor (KNN), Naïve Bayes, and support vector machine (SVM). Experimental results reveal that reducing the feature set from 2048 to 344 allows the ResNet-50-FTNCA-KNN model to achieve 100% accuracy, significantly enhancing AD detection. This approach will aid in the early diagnosis and treatment of AD patients.

Machine learning-based assessment of morphometric abnormalities distinguishes bipolar disorder and major depressive disorder.

Bipolar disorder (BD) and major depressive disorder (MDD) have overlapping clinical presentations which may make it difficult for clinicians to distinguish them potentially resulting in misdiagnosis. This study combined structural MRI and machine learning techniques to determine whether regional morphological differences could distinguish patients with BD and MDD. A total of 123 participants, including BD (n = 31), MDD (n = 48), and healthy controls (HC, n = 44), underwent high-resolution 3D T1-weighted imaging. Cortical thickness, surface area, and subcortical volumes were measured using FreeSurfer software. Common and classic machine learning models were utilized to identify distinct morphometric alterations between BD and MDD. Significant morphological differences were observed in both common and distinct brain regions between BD, MDD, and HC. Specifically, abnormalities in the amygdala, thalamus, medial orbitofrontal cortex and fusiform were observed in both BD and MDD compared with HC. Relative to HC, unique differences in BD were identified in the lateral occipital and inferior/middle temporal regions, whereas MDD exhibited differences in nucleus accumbens and middle temporal regions. BD exhibited larger surface area in right middle temporal gyrus and greater right nucleus accumbens volume compared to MDD. The integration of two-stage models, including deep neural network (DNN) and support vector machine (SVM), achieved an accuracy rate of 91.2% in discriminating individuals with BD from MDD. These findings demonstrate that structural MRI combined with machine learning techniques can accurately discriminate individuals with BD from MDD, and provide a foundation supporting the potential of this approach to improve diagnostic accuracy.

Are neurasthenia and depression the same disease entity? An electroencephalography study

BackgroundThe neurasthenia–depression controversy has lasted for several decades. It is challenging to solve the argument by symptoms alone for syndrome-based disease classification. Our aim was to identify objective electroencephalography (EEG) measures that can differentiate neurasthenia from major depressive disorder (MDD).MethodsBoth electronic medical information records and EEG records from patients with neurasthenia and MDD were gathered. The demographic and clinical characteristics, EEG power spectral density, and functional connectivity were compared between the neurasthenia and MDD groups. Machine Learning methods such as random forest, logistic regression, support vector machines, and k nearest neighbors were also used for classification between groups to extend the identification that there is a significant different pattern between neurasthenia and MDD.ResultsWe analyzed 305 patients with neurasthenia and 45 patients with MDD. Compared with the MDD group, patients with neurasthenia reported more somatic symptoms and less emotional symptoms (p < 0.05). Moreover, lower theta connectivity was observed in patients with neurasthenia compared to those with MDD (p < 0.01). Among the classification models, random forest performed best with an accuracy of 0.93, area under the receiver operating characteristic curve of 0.97, and area under the precision-recall curve of 0.96. The essential feature contributing to the model was the theta connectivity.LimitationsThis is a retrospective study, and medical records may not include all the details of a patient’s syndrome. The sample size of the MDD group was smaller than that of the neurasthenia group.ConclusionNeurasthenia and MDD are different not only in symptoms but also in brain activities.

Hyperspectral Band Selection with Unique Pixel Extraction and Adaptive Neighbor Clustering

Band selection is an effective way to reduce redundant information, while preserving the physical properties of hyperspectral images (HSI). However, most band selection methods merely consider the relevance and separability between pairs of bands and ignore those for different ground objects. To solve these issues, we propose a Unique Pixel extraction and Adaptive Neighbor Clustering (UPANC) band selection method in this theoretical study. First, in consideration of the characteristics of HSI data and tasks, unique pixels are obtained with a low-rank representation, where the importance of bands is analyzed from both spectral and spatial perspectives. Second, an adaptive neighbor clustering method is designed based on the unique pixels, which groups bands into several clusters through optimizing the graph structure under label smoothness. With support vector machines (SVM) as the classifier, the UPANC method achieved good performance, where the overall accuracy scores were 89.05%, 82.62%, and 92.07% on the Houston, IndianPines, and Pavia University datasets, respectively. The experimental results illustrated the advantages of the UPANC method, which could select optimal bands to enhance the performance in land cover observation.

Deep neural network and transfer learning for annual wetland vegetation mapping using sentinel-2 time-series data in the heterogeneous lake floodplain environment

ABSTRACT The lake-floodplain wetlands are characterized by high biodiversity, difficult access, and significant environmental changes. Traditional remote sensing mapping methods struggle to generate consistent time-series data on wetland vegetation communities. Current research has endeavoured to address this issue through the application of deep learning methodologies. However, a significant limitation of these models is their reliance on a substantial volume of training samples, which contradicts the difficulty and high cost of obtaining samples from the lake-floodplain wetlands. Whether it is possible to construct a transferable deep learning model under small sample conditions and apply it to the mapping of lake-floodplain wetlands is an urgent issue that needs to be addressed. To solve this problem, this study first constructed a deep neural network (DNN) designed specifically for mapping complex lake-floodplain wetland vegetation under conditions of limited sample size. Subsequently, using 2021 as a reference year, a novel histogram threshold method was proposed to identify the unchanged samples for the target transfer years of 2019, 2020, 2022, and 2023. Finally, annual wetland vegetation mapping was performed in Poyang Lake using DNN and sample transfer learning (STL). The results showed that high-quality annual time-series data of wetland vegetation can be generated using the constructed DNN and STL, with all overall accuracies exceeding 80%. The histogram threshold method, which combines SAD and NDVI indicators of key phenological period, can effectively solve the problem of difficulty in determining the unchanged samples in transfer learning for heterogeneous lake wetlands. Furthermore, the performance of STL based on the constructed DNN model was significantly superior to those based on support vector machine and random forest algorithms for mapping annual wetland vegetation communities using limited training samples. This study demonstrates that the effective application of DNN and STL will be highly beneficial for long-term monitoring of vegetation in lake-floodplain wetlands, particularly where sample availability is limited.

Sentiment Analysis of Free Online Novel Applications Using the Support Vector Machine Method

Sentiment analysis is a study to analyze opinions and perceptions of various topics, products, or services. With the advancement of technology, people now have easier access to literary works online, including novels. The shift from offline to online reading has resulted in a large volume of review data, necessitating an automated system to classify this data. This research aims to analyze the sentiment of reviews for online novel applications using the Support Vector Machine (SVM) algorithm. The data used in this study was gathered from user reviews of the Wattpad, Noveltoon, and Joylada applications downloaded from the Google Play Store. The results show that the Wattpad application achieved 63% accuracy, 50% precision, 64% recall, and 56% F1-score, with a 41% positive and 59% negative sentiment distribution. The Noveltoon application achieved 70% accuracy, 69% precision, 73% recall, and 71% F1 score, with a 48% positive and 52% negative sentiment distribution. The Joylada application recorded 67% accuracy, 55% precision, 92% recall, and 69% F1-score, with a 57% positive and 43% negative sentiment distribution. The results of this analysis can help understand user preferences towards online novel applications and provide insights into their impact on the application's image and user interactions.

Characterizing brain network alterations in cervical spondylotic myelopathy using static and dynamic functional network connectivity and machine learning.

Cervical spondylotic myelopathy (CSM) is a debilitating condition that affects the cervical spine, leading to neurological impairments. While the neural mechanisms underlying CSM remain poorly understood, changes in brain network connectivity, particularly within the context of static and dynamic functional network connectivity (sFNC and dFNC), may provide valuable insights into disease pathophysiology. This study investigates brain-wide connectivity alterations in CSM patients using both sFNC and dFNC, combined with machine learning approaches, to explore their potential as biomarkers for disease classification and progression. A total of 191 participants were included in this study, comprising 108 CSM patients and 83 healthy controls (HCs). Resting-state fMRI data were used to derive functional connectivity networks (FCNs), which were further analyzed to obtain sFNC and dFNC features. K-means clustering was applied to identify distinct dFNC states, and machine learning models, including support vector machine (SVM), decision tree (DT), linear discriminant analysis (LDA), logistic regression (LR), and random forests (RF), were constructed to classify CSM patients and HCs based on FNC features. The sFNC analysis revealed significant alterations in brain network connectivity in CSM patients, including enhanced connectivity between the posterior default mode network (pDMN) and ventral attention network (vAN), and between the right and left frontoparietal networks (rFPN and lFPN), alongside weakened connectivity in multiple other network pairs. K-means clustering of dFNC identified four distinct functional states, with CSM patients exhibiting altered connectivity in State 1 and State 3. Machine learning models based on sFNC demonstrated excellent classification performance, with the SVM model achieving an AUC of 0.92, accuracy of 85.86%, and sensitivity and specificity both exceeding 0.80. Models based on dFNC also performed well, with the State 3-based model yielding an AUC of 0.91 and accuracy of 84.97%. Our findings highlight significant alterations in both sFNC and dFNC in CSM patients, suggesting that these connectivity changes may reflect underlying neural mechanisms of the disease. Machine learning models based on FNC features, particularly SVM, exhibit strong potential for classifying CSM patients and may serve as valuable neuroimaging biomarkers for diagnosis and monitoring disease progression. Future research should explore longitudinal studies and multimodal neuroimaging approaches to further validate these findings.

Using Inertial Measurement Units and Machine Learning to Classify Body Positions of Adults in a Hospital Bed

In hospitals, timely interventions can prevent avoidable clinical deterioration. Early recognition of deterioration is vital to stopping further decline. Measuring the way patients position themselves in bed and change their positions may signal when further assessment is necessary. While inertial measurement units (IMUs) have been used in health research, their use inside hospitals has been limited. This study explores the use of IMUs with machine learning to continuously capture, classify and visualise patient positions in hospital beds. The participants attended a data collection session in a simulated hospital bedspace and were asked to adopt nine positions. Movement data were captured using five IMU Xsens DOTs attached to the forehead, wrists and ankles. Support Vector Machine (SVM) and K-Nearest Neighbours classifiers were trained using five different combinations of sensors (e.g., right wrist only, right and left wrist) to determine body positions. Data from 30 participants were analysed. The highest accuracy (87.7%) was achieved by SVM using forehead and wrist sensors. Adding data from ankle sensors reduced the accuracy. To preserve patient privacy in a hospital setting, a 3D visualisation was developed in Unity, offering a non-identifiable representation of patient positions. This system could help clinicians monitor changes in position which may signal clinical deterioration.

The Role of Machine Learning Approaches in Pediatric Oncology: A Systematic Review.

To enhance patient outcomes in pediatric cancer, a better understanding of the medical and biologicalrisk variables is required. With the growing amount of data accessible to research in pediatric cancer, machine learning (ML) is a form of algorithmic inference from sophisticated statistical techniques. In addition to highlighting developments and prospects in the field, the objective of this systematic study was to methodically describe the state of ML in pediatric oncology. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to search for relevant studies on four distinct databases (Scopus, Web of Science, PubMed, and Cochrane Library). A total of 1536 relevant studies were retrieved to the EndNote library(Clarivate, Philadelphia, USA) where duplicates were removed and the rest of the studies were assessed for eligibility based on titles, abstracts, and the availability of full-text articles. After assessing the studies for eligibility, we found 42 studies eligible for inclusion in this systematic review. We found nine studies on liquid tumors, 13 on solid tumors, and 20 on central nervous system (CNS) tumors. ML goals included classification, treatment response prediction, and dose optimization. Neural networks, k-nearest neighbors, random forests, support vector machines, and naive Bayeswere among the techniques employed. The identified studies' strengths included treatment response prediction and automated analysis that matched or outperformed physician comparators. Significant variation in clinical applicability,criteria for reporting, limited sample numbers, and the absence of external validation cohorts were among the common issues. We found places where MLcan improve clinical care in manners that would not be possible otherwise. Even though MLhas great promise for enhancing pediatric cancer diagnosis, decision-making, and monitoring, the discipline is still in its infancy, and standards and recommendations will support future research to guarantee robust methodologic design and maximize therapeutic applicability.

Establishment and Validation of the Diagnostic Value of Oligodendrocyte-related Genes in Alzheimer's Disease.

AD is a demyelinating disease. Myelin damage initiates the pathological process of AD, resulting in abnormal synaptic function and cognitive decline. The myelin sheath formed by oligodendrocytes (OL) is a crucial component of white matter. Investigating AD from the perspective of OL may offer novel diagnostic and therapeutic perspectives. This study aimed to analyze the association between OL-related genes and Alzheimer's disease (AD) using bioinformatics and verify this association via molecular biology experiments. The AD datasets were acquired from the Gene Expression Omnibus (GEO) database of NCBI. Consensus clustering was employed to determine the subtypes of AD, followed by evaluating the clinical characteristics of these subtypes. Subsequently, immune infiltration analysis of relevant genes and Weighted Gene Co-expression Network Analysis (WGCNA) were conducted to identify modules and hub genes associated with AD progression. The intersection of genes obtained was analyzed using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. To narrow down the scope and identify OL-related genes with diagnostic potential, three machine learning algorithms were utilized. In addition, the eXtreme Sum (XSum) algorithm was used to screen small molecule drug candidates based on the connectivity map (CMAP) database. Finally, these identified genes were validated using Real-time fluorescence quantitative PCR (RT-qPCR). Among the three subtypes of AD, Cluster A and Cluster C exhibited increased levels of Braak and neurofibrillary tangles compared to Cluster B. The proportion of females was greater than that of males among the three subclasses of AD. There were no significant differences in age among the three subclasses, but significant differences in gene expression existed. Through WGCNA analysis, 108 genes were identified. Among these, 16 genes were identified as shared genes by the least absolute shrinkage and selection operator (LASSO), random forest (RF), and support vector machines (SVM) algorithms, and logistic regression further determined 11 genes. The establishment of a nomogram demonstrated the significance of these 11 genes in AD. The "XSum" algorithm revealed five drugs with therapeutic potential for AD. qPCR analysis revealed the upregulation and downregulation of the highlighted genes. According to this study, 11 genes related to OL were also found to be associated with immune cell infiltration in AD patients. These genes demonstrated potential diagnostic value for AD. Additionally, we screened five small molecular drugs that exhibit potential therapeutic effects on AD. This research provides a new perspective for personalized clinical management and treatment of AD.

Generalisation of EEG-Based Pain Biomarker Classification for Predicting Central Neuropathic Pain in Subacute Spinal Cord Injury

Background: The objective was to test the generalisability of electroencephalography (EEG) markers of future pain using two independent datasets. Methods: Datasets, A [N = 20] and B [N = 35], were collected from participants with subacute spinal cord injury who did not have neuropathic pain at the time of recording. In both datasets, some participants developed pain within six months, (PDP) will others did not (PNP). EEG features were extracted based on either band power or Higuchi fractal dimension (HFD). Three levels of generalisability were tested: (1) classification PDP vs. PNP in datasets A and B separately; (2) classification between groups in datasets A and B together; and (3) classification where one dataset (A or B) was used for training and testing, and the other for validation. A novel normalisation method was applied to HFD features. Results: Training and testing of individual datasets achieved classification accuracies of &gt;80% using either feature set, and classification of joint datasets (A and B) achieved a maximum accuracy of 86.4% (HFD, support vector machine (SVM)). With normalisation and feature reduction (principal components), the validation accuracy was 66.6%. Conclusions: An SVM classifier with HFD features showed the best robustness, and normalisation improved the accuracy of predicting future neuropathic pain well above the chance level.

Advancing COVID-19 Treatment: The Role of Non-covalent Inhibitors Unveiled by Integrated Machine Learning and Network Pharmacology.

The COVID-19 pandemic has necessitated rapid advancements in therapeutic discovery. This study presents an integrated approach combining machine learning (ML) and network pharmacology to identify potential non-covalent inhibitors against pivotal proteins in COVID-19 pathogenesis, specifically B-cell lymphoma 2 (BCL2) and Epidermal Growth Factor Receptor (EGFR). Employing a dataset of 13,107 compounds, ML algorithms such as k-Nearest Neighbors (kNN), Support Vector Machine (SVM), Random Forest (RF), and Naïve Bayes (NB) were utilized for screening and predicting active inhibitors based on molecular features. Molecular docking and molecular dynamics simulations, conducted over a 100 nanosecond period, enhanced the ML-based screening by providing insights into the binding affinities and interaction dynamics with BCL2 and EGFR. Network pharmacology analysis identified these proteins as hub targets within the COVID-19 protein-protein interaction network, highlighting their roles in apoptosis regulation and cellular signaling. The identified inhibitors exhibited strong binding affinities, suggesting potential efficacy in disrupting viral life cycles and impeding disease progression. Comparative analysis with existing literature affirmed the relevance of BCL2 and EGFR in COVID-19 therapy and underscored the novelty of integrating network pharmacology with ML. This multidisciplinary approach establishes a framework for emerging pathogen treatments and advocates for subsequent in vitro and in vivo validation, emphasizing a multi-targeted drug design strategy against viral adaptability. This study's findings are crucial for the ongoing development of therapeutic agents against COVID-19, leveraging computational and network-based strategies.

Predictive modeling of transverse cracking in continuously reinforced concrete pavement: a machine learning approach

Abstract Accurate prediction of transverse cracking in Continuously Reinforced Concrete Pavement (CRCP) is critical for improving infrastructure management procedures and preserving the road network's long-term durability and safety. This paper conducts a thorough analysis into predicting transverse cracking in CRCP using machine learning approaches. The research involved meticulous data preparation, feature selection, and evaluation of various machine learning models to identify the most effective predictor. Key variables such as pavement age, total thickness, temperature, freeze index, traffic volume, precipitation, and initial International Roughness Index (IRI) were analyzed for their impact on transverse cracking occurrences. Sensitivity analysis was conducted to assess the influence of individual input variables on model predictions. Results indicated that the cubic Support Vector Machine (SVM) model outperformed other models, demonstrating exceptional predictive accuracy. Furthermore, sensitivity analysis revealed significant correlations between input variables and transverse cracking occurrences, emphasizing the importance of considering a holistic range of factors in pavement engineering and maintenance strategies. Our findings, which provide insights into the intricate interactions between input factors and pavement distress, help to create tailored treatments and methods for minimizing transverse cracking and enhancing CRCP performance.&amp;#xD;

Enhancing Small-for-Gestational-Age Prediction: Multi-Country Validation of Nuchal Thickness, Estimated Fetal Weight, and Machine Learning Models.

The first objective is to develop a nuchal thickness reference chart. The second objective is to compare rule-based algorithms and machine learning models in predicting small-for-gestational-age infants. This retrospective study involved singleton pregnancies at University Malaya Medical Centre, Malaysia, developed a nuchal thickness chart and evaluated its predictive value for small-for-gestational-age using Malaysian and Singapore cohorts. Predictive performance using conjunctive (AND)/disjunctive (OR) rule-based algorithms was assessed. Seven machine learning models were trained on Malaysia data and evaluated on both Malaysia and Singapore cohorts. 5519 samples were collected from the University Malaya Medical Centre. Small-for-gestational-age infants exhibit significantly lower nuchal thickness (small-for-gestational-age: 4.57 [1.04] mm, appropriate-for-gestational-age: 4.86 [1.06] mm, p<0.001). Implementing disjunctive rule (nuchal thickness<10th centile or estimated fetal weight<10th centile) significantly improved small-for-gestational-age prediction across all growth charts, with balanced accuracy gains of 5.83% in Malaysia (p<0.05) and 7.75% in Singapore. The best model for predicting small-for-gestational-age was: logistic regression with five variables (abdominal circumference, femur length, nuchal thickness, maternal age, and ultrasound-confirmed gestational age), which achieved an area under the curve of 0.75 for Malaysia cohorts; support vector machine with all variables, achieved area under the curve of 0.81 for Singapore cohorts. Small-for-gestational-age infants demonstrate significantly reduced second-trimester nuchal thickness. Employing the disjunctive rule enhanced small-for-gestational-age prediction. Logistic regression and support vector machines show superior performance among all models, highlighting the advantages of machine learning. Larger prospective studies are needed to assess clinical utility.

Predicting Heart Failure: A Comparative Approach between Artificial Neural Networks and Support Vector Machines

Predicting Heart Failure: A Comparative Approach between Artificial Neural Networks and Support Vector Machines

Continuous wavelet transform based face milling tool condition classification using support vector machine and K-star algorithm–a comparative study

Continuous wavelet transform based face milling tool condition classification using support vector machine and K-star algorithm–a comparative study