Feature Selection Research Articles

Using feature selection methods to identify important factors of in-game consumption

ABSTRACT In-game consumption prediction remains under-researched. How to design products that satisfy the needs of users and attract more consumers has also become one of important issues for game App developers. Therefore, this study aims to integrate Technology Acceptance Model into Innovation Theory Diffusion (IDT) to define potential factors that influence users to make in-app purchases. Support Vector Machine -Recursive Feature Elimination (SVM-RFE), Decision Trees (DT), and Information Gain (IG) were utilized to identify the key factors. Perceived usefulness, perceived ease of use, result demonstrability, mass media, perceived enjoyment, cognitive concentration, use context are identified as important factors for in-game purchase.

A Machine Learning-Based Prediction Model for Acute Kidney Injury in Patients With Community-Acquired Pneumonia: Multicenter Validation Study.

Acute kidney injury (AKI) is common in patients with community-acquired pneumonia (CAP) and is associated with increased morbidity and mortality. This study aimed to establish and validate predictive models for AKI in hospitalized patients with CAP based on machine learning algorithms. We trained and externally validated 5 machine learning algorithms, including logistic regression, support vector machine, random forest, extreme gradient boosting, and deep forest (DF). Feature selection was conducted using the sliding window forward feature selection technique. Shapley additive explanations and local interpretable model-agnostic explanation techniques were applied to the optimal model for visual interpretation. A total of 6371 patients with CAP met the inclusion criteria. The development of CAP-associated AKI (CAP-AKI) was recognized in 1006 (15.8%) patients. The 11 selected indicators were sex, temperature, breathing rate, diastolic blood pressure, C-reactive protein, albumin, white blood cell, hemoglobin, platelet, blood urea nitrogen, and neutrophil count. The DF model achieved the best area under the receiver operating characteristic curve (AUC) and accuracy in the internal (AUC=0.89, accuracy=0.90) and external validation sets (AUC=0.87, accuracy=0.83). Furthermore, the DF model had the best calibration among all models. In addition, a web-based prediction platform was developed to predict CAP-AKI. The model described in this study is the first multicenter-validated AKI prediction model that accurately predicts CAP-AKI during hospitalization. The web-based prediction platform embedded with the DF model serves as a user-friendly tool for early identification of high-risk patients.

Sound insulation prediction and optimization of wooden support structure for high-speed train floor based on machine learning

In order to improve the sound insulation performance of high-speed train floors, this study first obtained the necessary data for model training based on the reverberation test method, and then conducted data sorting and feature selection. Next, the maximum mutual information minimum redundancy (mRMR) feature selection algorithm was used to calculate the selected features and screen out a subset of significant features. Subsequently, the decision tree, BP neural network, and support vector machine regression (SVR) methods were applied in sequence, and the standardized feature data were used for the high-speed train floor under the same evaluation criteria of the mean square error (MSE) and coefficient of determination (R2). We conducted training and validation of the sound insulation prediction models for timber-framed support structures. The prediction accuracy of the trained model was compared and evaluated with the finite element statistical energy analysis (FE-SEA) prediction model. Finally, the SVR model was used to optimize the design under constraint conditions. The research results show that based on the research object, sample library, and model training in this article, compared with the FE-SEA model, the prediction error of the SVR model is only 0.3 dB, showing better performance. In engineering practice, the SVR model can effectively optimize the wooden support structure in the floor under certain constraints, and it predicts that the weighted sound insulation of the entire floor is 50.45 dB, which has important engineering application value.

XGBoost-SHAP-based interpretable diagnostic framework for knee osteoarthritis: a population-based retrospective cohort study

ObjectiveTo use routine demographic and clinical data to develop an interpretable individual-level machine learning (ML) model to diagnose knee osteoarthritis (KOA) and to identify highly ranked features.MethodsIn this retrospective, population-based cohort study, anonymized questionnaire data was retrieved from the Wu Chuan KOA Study, Inner Mongolia, China. After feature selections, participants were divided in a 7:3 ratio into training and test sets. Class balancing was applied to the training set for data augmentation. Four ML classifiers were compared by cross-validation within the training set and their performance was further analyzed with an unseen test set. Classifications were evaluated using sensitivity, specificity, positive predictive value, negative predictive value, accuracy, area under the curve(AUC), G-means, and F1 scores. The best model was explained using Shapley values to extract highly ranked features.ResultsA total of 1188 participants were investigated in this study, among whom 26.3% were diagnosed with KOA. Comparatively, XGBoost with Boruta exhibited the highest classification performance among the four models, with an AUC of 0.758, G-means of 0.800, and F1 scores of 0.703. The SHAP method reveals the top 17 features of KOA according to the importance ranking, and the average of the experience of joint pain was recognized as the most important features.ConclusionsOur study highlights the usefulness of machine learning in unveiling important factors that influence the diagnosis of KOA to guide new prevention strategies. Further work is needed to validate this approach.

A Hierarchical Machine Learning-Based Strategy for Mapping Grassland in Manitoba’s Diverse Ecoregions

Accurate and reliable knowledge about grassland distribution is essential for farmers, stakeholders, and government to effectively manage grassland resources from agro-economical and ecological perspectives. This study developed a novel pixel-based grassland classification approach using three supervised machine learning (ML) algorithms, which were assessed in the province of Manitoba, Canada. The grassland classification process involved three stages: (1) to distinguish between vegetation and non-vegetation covers, (2) to differentiate grassland from non-grassland landscapes, and (3) to identify three specific grassland classes (tame, native, and mixed grasses). Initially, this study investigated different satellite data, such as Sentinel-1 (S1), Sentinel-2 (S2), and Landsat 8 and 9, individually and combined, using the random forest (RF) method, with the best performance at the first two steps achieved using a combination of S1 and S2. The combination was then utilized to conduct the first two steps of classification using support vector machine (SVM) and gradient tree boosting (GTB). In step 3, after filtering out non-grassland pixels, the performance of RF, SVM, and GTB classifiers was evaluated with combined S1 and S2 data to distinguish different grassland types. Eighty-nine multitemporal raster-based variables, including spectral bands, SAR backscatters, and digital elevation models (DEM), were input for ML models. RF had the highest classification accuracy at 69.96% overall accuracy (OA) and a Kappa value of 0.55. After feature selection, the variables were reduced to 61, increasing OA to 72.62% with a Kappa value of 0.58. GTB ranked second, with its OA and Kappa values improving from 67.69% and 0.50 to 72.18% and 0.58 after feature selection. The impact of raster data quality on grassland classification accuracy was assessed through multisensor image fusion. Grassland classification using the Hue, Saturation, and Value (HSV) fused images showed higher OA (59.18%) and Kappa values (0.36) than the Brovey Transform (BT) and non-fused images. Finally, a web map was created to show grassland results within the Soil Landscapes of Canada (SLC) polygons, relating soil landscapes to grassland distribution and providing valuable information for decision-makers and researchers. Future work may include extending the current methodology by considering other influential variables, like meteorological parameters or soil properties, to create a comprehensive grassland inventory across the whole Prairie ecozone of Canada.

Synaptic basis of feature selectivity in hippocampal neurons.

A central question in neuroscience is how synaptic plasticity shapes the feature selectivity of neurons in behaving animals1. Hippocampal CA1 pyramidal neurons display one of the most striking forms of feature selectivity by forming spatially and contextually selective receptive fields called place fields, which serve as a model for studying the synaptic basis of learning and memory. Various forms of synaptic plasticity have been proposed as cellular substrates for the emergence of place fields. However, despite decades of work, our understanding of how synaptic plasticity underlies place-field formation and memory encoding remains limited, largely due to a shortage of tools and technical challenges associated with the visualization of synaptic plasticity at the single-neuron resolution in awake behaving animals. Toaddress this, we developed an all-optical approach to monitor the spatiotemporal tuning and synaptic weight changes of dendritic spines before and after the induction of a place field in single CA1 pyramidal neurons during spatial navigation. We identified a temporally asymmetric synaptic plasticity kernel resulting from bidirectional modifications of synaptic weights around the induction of a place field. Our work identified compartment-specific differences in the magnitude and temporal expression of synaptic plasticity between basal dendritesand oblique dendrites. Our results provide experimental evidence linking synaptic plasticity to the rapid emergence of spatial selectivity in hippocampal neurons, a critical prerequisite for episodic memory.

Biomarker discovery and development of prognostic prediction model using metabolomic panel in breast cancer patients: a hybrid methodology integrating machine learning and explainable artificial intelligence

BackgroundBreast cancer (BC) is a significant cause of morbidity and mortality in women. Although the important role of metabolism in the molecular pathogenesis of BC is known, there is still a need for robust metabolomic biomarkers and predictive models that will enable the detection and prognosis of BC. This study aims to identify targeted metabolomic biomarker candidates based on explainable artificial intelligence (XAI) for the specific detection of BC.MethodsData obtained after targeted metabolomics analyses using plasma samples from BC patients (n = 102) and healthy controls (n = 99) were used. Machine learning (ML) models based on raw data were developed, then feature selection methods were applied, and the results were compared. SHapley Additive exPlanations (SHAP), an XAI method, was used to clinically explain the decisions of the optimal model in BC prediction.ResultsThe results revealed that variable selection increased the performance of ML models in BC classification, and the optimal model was obtained with the logistic regression (LR) classifier after support vector machine (SVM)-SHAP-based feature selection. SHAP annotations of the LR model revealed that Leucine, isoleucine, L-alloisoleucine, norleucine, and homoserine acids were the most important potential BC diagnostic biomarkers. Combining the identified metabolite markers provided robust BC classification measures with precision, recall, and specificity of 89.50%, 88.38%, and 83.67%, respectively.ConclusionIn conclusion, this study adds valuable information to the discovery of BC biomarkers and underscores the potential of targeted metabolomics-based diagnostic advances in the management of BC.

Machine learning-driven simplification of the hypomania checklist-32 for adolescent: a feature selection approach

BackgroundThe Hypomania Checklist-32 is widely used to screen for bipolar disorder, but its length can be challenging for adolescents with manic symptoms. This study aimed to develop a shortened version of the HCL-32 tailored for adolescents using machine learning techniques.MethodsData from 2,850 adolescents (mean age 15.50 years, 68.81% female) who completed the HCL-32 were analyzed. Random forest (RF) and gradient boosting machine (GBM) algorithms were employed for feature selection. The area under the curve (AUC) was used to evaluate model performance. Receiver operating characteristic (ROC) analysis was conducted to determine optimal cutoff points for the shortened scale.ResultsAn 8-item version of the HCL-32 was derived, maintaining high predictive accuracy (AUC = 0.97). The selected items captured core symptoms of adolescent mania, including increased energy, risk-taking, and irritability. Two cutoff points were identified: a score of 3 offered high specificity (0.98) and positive predictive value (0.98), while a score of 4 provided balanced sensitivity (0.87) and specificity (0.94) with the highest overall accuracy (0.91).ConclusionsThe machine learning-driven 8-item version of the HCL-32 demonstrates strong diagnostic utility for adolescent bipolar disorder, offering a more efficient screening tool without sacrificing clinical sensitivity. This shortened scale may improve assessment feasibility and accuracy in clinical settings, addressing the unique challenges of diagnosing bipolar disorder in adolescents.

Simultaneous Clustering and Feature Selection using Social Group Optimization with Dynamic Threshold Setting for Microarray Data

Simultaneous Clustering and Feature Selection using Social Group Optimization with Dynamic Threshold Setting for Microarray Data

Novel neural network classification of maternal fetal ultrasound planes through optimized feature selection

Ultrasound (US) imaging is an essential diagnostic technique in prenatal care, enabling enhanced surveillance of fetal growth and development. Fetal ultrasonography standard planes are crucial for evaluating fetal development parameters and detecting abnormalities. Real-time imaging, low cost, non-invasiveness, and accessibility make US imaging indispensable in clinical practice. However, acquiring fetal US planes with correct fetal anatomical features is a difficult and time-consuming task, even for experienced sonographers. Medical imaging using AI shows promise for addressing current challenges. In response to this challenge, a Deep Learning (DL)-based automated categorization method for maternal fetal US planes are introduced to enhance detection efficiency and diagnosis accuracy. This paper presents a hybrid optimization technique for feature selection and introduces a novel Radial Basis Function Neural Network (RBFNN) for reliable maternal fetal US plane classification. A large dataset of maternal–fetal screening US images was collected from publicly available sources and categorized into six groups: the four fetal anatomical planes, the mother's cervix, and an additional category. Feature extraction is performed using Gray-Level Co-occurrence Matrix (GLCM), and optimization methods such as Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), and a hybrid Particle Swarm Optimization and Grey Wolf Optimization (PSOGWO) approach are utilized to select the most relevant features. The optimized features from each algorithm are then input into both conventional and proposed DL models. Experimental results indicate that the proposed approach surpasses conventional DL models in performance. Furthermore, the proposed model is evaluated against previously published models, showcasing its superior classification accuracy. In conclusion, our proposed approach provides a solid foundation for automating the classification of fetal US planes, leveraging optimization and DL techniques to enhance prenatal diagnosis and care.

Radiomics and deep learning features of pericoronary adipose tissue on non-contrast computerized tomography for predicting non-calcified plaques

Background Inflammation of coronary arterial plaque is considered a key factor in the development of coronary heart disease. Early the plaque detection and timely treatment of the atherosclerosis could effectively reduce the risk of cardiovascular events. However, there is no study combining radiomics with deep learning techniques to predict non-calcified plaques (NCP) in coronary artery at present. Objective To investigate the value of combination of radiomics and deep learning features based on non-contrast computerized tomography (CT) scans of pericoronary adipose tissue (PCAT), integrating with clinical risk factors of patients, in identifying coronary inflammation and predicting the presence of NCP. Methods The clinical and imaging data of 353 patients were analyzed. The region of interest (ROI) of PCAT was manually outlined on non-contrast CT scan images, like coronary CT calcium score sequential images, then the radiomics and deep learning features in ROIs were extracted respectively. In training set (Center 1), after performing feature selection, radiomics and deep learning feature models were established, meanwhile, clinical models were built. Finally, combined models were developed out via integrating clinical, radiomics, and deep learning features. The predictive performance of the four feature model groups (clinical, radiomics, deep learning, and three combination) was assessed by seven different machine learning models through generation of receiver operating characteristic curves (ROC) and the calculation of area under the curve (AUC), sensitivity, specificity, and accuracy. Furthermore, the predictive performance of each model was validated in an external validation set (Center 2). Results For the single model comparation, eXtreme Gradient Boosting (XGBoost) showed the best performance among the clinical model group in the validation set. And Random Forest (RF) exhibited the best indicative performance not only among the radiomics feature group but also in the deep learning feature model group. What's more, among the combined model group, RF still displayed the best predictive performance, with the value of AUC, sensitivity, specificity, and accuracy in the validation set are 0.963, 0.857, 0.929, and 0.905. Conclusion The RF model in the combined model group based on non-contrast CT scan PCAT can predict the presence of NCP more accurately and has the potential for preliminary screening of the NCP.

Integrative Stacking Machine Learning Model for Small Cell Lung Cancer Prediction Using Metabolomics Profiling

Background: Small cell lung cancer (SCLC) is an extremely aggressive form of lung cancer, characterized by rapid progression and poor survival rates. Despite the importance of early diagnosis, the current diagnostic techniques are invasive and restricted. Methods: This study presents a novel stacking-based ensemble machine learning approach for classifying small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) using metabolomics data. The analysis included 191 SCLC cases, 173 NSCLC cases, and 97 healthy controls. Feature selection techniques identified significant metabolites, with positive ions proving more relevant. Results: For multi-class classification (control, SCLC, NSCLC), the stacking ensemble achieved 85.03% accuracy and 92.47 AUC using Support Vector Machine (SVM). Binary classification (SCLC vs. NSCLC) further improved performance, with ExtraTreesClassifier reaching 88.19% accuracy and 92.65 AUC. SHapley Additive exPlanations (SHAP) analysis revealed key metabolites like benzoic acid, DL-lactate, and L-arginine as significant predictors. Conclusions: The stacking ensemble approach effectively leverages multiple classifiers to enhance overall predictive performance. The proposed model effectively captures the complementary strengths of different classifiers, enhancing the detection of SCLC and NSCLC. This work accentuates the potential of combining metabolomics with advanced machine learning for non-invasive early lung cancer subtype detection, offering an alternative to conventional biopsy methods.

Leveraging machine learning models for anemia severity detection among pregnant women following ANC: Ethiopian context

BackgroundAnemia during pregnancy is a significant public health concern, particularly in resource-limited settings. Machine learning (ML) offers promising avenues for improved anemia detection and management. This study investigates the potential of ML models in predicting anemia severity among pregnant women attending Antenatal Care (ANC) visits in Ethiopia.MethodsData from the Ethiopian Demographic Health Survey, specialized hospitals, and public hospitals were utilized. The dataset included individuals diagnosed with severe (65.12%), moderate (15.63%), mild (16.65%) anemia, and non-anemic (2.61%) cases. Feature selection employed filter methods based on mutual information, and F-score was used to assess anemia severity prediction across four classes. Six ML models (MLP-NN, XGBoost, GNB, Decision Tree, Random Forest, and KNN) were evaluated using accuracy, precision, recall, and F1-score.ResultsThe Random Forest classifier achieved the best overall performance across all categories, with an accuracy of 97%, precision of 93%, recall of 93%, and F1-score of 93%. This indicates high true positive rates and low false positive rates. While other models like XGBoost, MLP-NN, and Decision Tree showed good performance, they weren’t quite as strong as Random Forest. Classifiers like KNN and GNB had lower overall accuracy and a tendency to misclassify some cases.ConclusionsThis study demonstrates the promising potential of Random Forest in predicting anemia severity among pregnant women in Ethiopia. The findings contribute to a more holistic understanding of anemia risk factors and pave the way for improved early detection and targeted interventions.

Intelligent Lung Cancer Detection: A Genetic Algorithm Machine Learning Fusion

Early detection of lung cancer is crucial for improving patient outcomes. While Deep Learning (DL) models have shown high accuracy in lung cancer diagnosis, they often require substantial computational resources. This study proposes a novel approach that leverages Genetic Algorithm (GA) to optimize feature selection and dimensionality reduction from lung cancer images. By integrating GA with conventional Machine Learning (ML) models, we demonstrate improved classification accuracy while minimizing computational requirements. Our experimental results show that combining GA with a feed-forward neural network classifier yields exceptional performance, achieving a classification accuracy of 99.70%. This approach offers a promising alternative to DL models for lung cancer detection, particularly in resource-constrained settings.

Uncovering the Diagnostic Power of Radiomic Feature Significance in Automated Lung Cancer Detection: An Integrative Analysis of Texture, Shape, and Intensity Contributions

Background: Lung cancer still maintains the leading position among causes of death in the world; the process of early detection surely contributes to changes in the survival of patients. Standard diagnostic methods are grossly insensitive, especially in the early stages. In this paper, radiomic features are discussed that can assure improved diagnostic accuracy through automated lung cancer detection by considering the important feature categories, such as texture, shape, and intensity, originating from the CT DICOM images. Methods: We developed and compared the performance of two machine learning models—DenseNet-201 CNN and XGBoost—trained on radiomic features with the ability to identify malignant tumors from benign ones. Feature importance was analyzed using SHAP and techniques of permutation importance that enhance both the global and case-specific interpretability of the models. Results: A few features that reflect tumor heterogeneity and morphology include GLCM Entropy, shape compactness, and surface-area-to-volume ratio. These performed excellently in diagnosis, with DenseNet-201 producing an accuracy of 92.4% and XGBoost at 89.7%. The analysis of feature interpretability ascertains its potential in early detection and boosting diagnostic confidence. Conclusions: The current work identifies the most important radiomic features and quantifies their diagnostic significance through a properly conducted feature selection process reflecting stability analysis. This provides the blueprint for feature-driven model interpretability in clinical applications. Radiomics features have great value in the automated diagnosis of lung cancer, especially when combined with machine learning models. This might improve early detection and open personalized diagnostic strategies for precision oncology.

Hybrid bagging and boosting with SHAP based feature selection for enhanced predictive modeling in intrusion detection systems

The novelty and growing sophistication of cyber threats mean that high accuracy and interpretable machine learning models are needed more than ever before for Intrusion Detection and Prevention Systems. This study aims to solve this challenge by applying Explainable AI techniques, including Shapley Additive explanations feature selection, to improve model performance, robustness, and transparency. The method systematically employs different classifiers and proposes a new hybrid method called Hybrid Bagging-Boosting and Boosting on Residuals. Then, performance is taken in four steps: the multistep evaluation of hybrid ensemble learning methods for binary classification and fine-tuning of performance; feature selection using Shapley Additive explanations values retraining the hybrid model for better performance and reducing overfitting; the generalization of the proposed model for multiclass classification; and the evaluation using standard information metrics such as accuracy, precision, recall, and F1-score. Key results indicate that the proposed methods outperform state-of-the-art algorithms, achieving a peak accuracy of 98.47% and an F1 score of 96.19%. These improvements stem from advanced feature selection and resampling techniques, enhancing model accuracy and balancing precision and recall. Integrating Shapley Additive explanations-based feature selection with hybrid ensemble methods significantly boosts the predictive and explanatory power of Intrusion Detection and Prevention Systems, addressing common pitfalls in traditional cybersecurity models. This study paves the way for further research on statistical innovations to enhance Intrusion Detection and Prevention Systems performance.

Implementation of Chernobyl disaster optimizer based feature selection approach to predict software defects

Background Software Defect Prediction (SDP) enables developers to investigate unscrambled faults in the inaugural parts of the software progression mechanism. However, SDP faces the threat of high dimensionality. Feature selection (FS) selects the finest features while carefully discarding others. Several meta-heuristic algorithms, like Genetic Algorithm, Particle Swarm Optimization, Differential Evolution, and Ant Colony Optimization, have been used to develop defect prediction models. However, these models have drawbacks like high cost, local optima trap, lower convergence rate, and higher parameter tuning. This study applies an innovative FS technique (FSCOA) rooted in Chernobyl Disaster Optimizer (CDO) technique. The proposed procedure intends to unwrap the best features for a prediction model while minimizing errors. Methods The proposed FSCOA investigated twelve public NASA software datasets from the PROMISE archive on Decision Tree, K-nearest neighbor, Naive Bayes, and Quantitative Discriminant Analysis classifiers. Furthermore, the accuracy of the recommended FSCOA method was correlated with existing FS techniques, like FSDE, FSPSO, FSACO, and FSGA. The statistical merit of the proposed measure was verified using Friedman and Holm tests. Results The experiment indicated that the proposed FSCOA approach bettered the accuracy in majority of the instances and achieved an average rank of 1.75 among other studied FS approaches while applying the Friedman test. Furthermore, the Holm test showed that the p-value was lower than or equivalent to the value of α/(A-i), except for the FSCOA and FSGA and FSCOA and FSACO models. Conclusion The results illustrated the supremacy of the prospective FSCOA procedure over extant FS techniques with higher accuracy in almost all cases due to its advantages like enhanced accuracy, the ability to deal with convoluted, high-magnitude datasets not grounded in local optima, and a faster convergence rate. These advantages empower the suggested FSCOA method to overcome the challenges of the other studied FS techniques.

ReLMM: Reinforcement Learning Optimizes Feature Selection in Modeling Materials.

A challenge to materials discovery is the identification of the physical features that are most correlated to a given target material property without redundancy. Such variables necessarily comprise the optimal search domain in subsequent material design. Here, we introduce a reinforcement learning-based material model (ReLMM) as a tool for analyzing a given database in identifying a minimal or near minimal subset of physical features for the design of a material with a given target property. We aim for minimality in the selected subset with respect to its size─smaller being better─ while maintaining the desired accuracy of the prediction. We have shown, using synthetic multiscale data sets, that ReLMM can identify the relative importance of features, and thus help identify which should be selected across scales. In the context of semiconducting materials, ReLMM can be used to improve the prediction of the band gap by identifying which features should be selected in model building. For metal halide perovskites, ReLMM was seen to find a near minimal data set at least as well as, if not better than, state-of-the-art feature selection tools such as LASSO and XGBoost. We also found that our domain-science oriented approach can be used to uncover the hierarchical structure of a material from a database consisting of molecular-scale, mesoscale and device-scale features and labels in complementarity with an earlier hierarchical model called NestedAE.

Improving the Performance of Electrotactile Brain–Computer Interface Using Machine Learning Methods on Multi-Channel Features of Somatosensory Event-Related Potentials

Traditional tactile brain–computer interfaces (BCIs), particularly those based on steady-state somatosensory–evoked potentials, face challenges such as lower accuracy, reduced bit rates, and the need for spatially distant stimulation points. In contrast, using transient electrical stimuli offers a promising alternative for generating tactile BCI control signals: somatosensory event-related potentials (sERPs). This study aimed to optimize the performance of a novel electrotactile BCI by employing advanced feature extraction and machine learning techniques on sERP signals for the classification of users’ selective tactile attention. The experimental protocol involved ten healthy subjects performing a tactile attention task, with EEG signals recorded from five EEG channels over the sensory–motor cortex. We employed sequential forward selection (SFS) of features from temporal sERP waveforms of all EEG channels. We systematically tested classification performance using machine learning algorithms, including logistic regression, k-nearest neighbors, support vector machines, random forests, and artificial neural networks. We explored the effects of the number of stimuli required to obtain sERP features for classification and their influence on accuracy and information transfer rate. Our approach indicated significant improvements in classification accuracy compared to previous studies. We demonstrated that the number of stimuli for sERP generation can be reduced while increasing the information transfer rate without a statistically significant decrease in classification accuracy. In the case of the support vector machine classifier, we achieved a mean accuracy over 90% for 10 electrical stimuli, while for 6 stimuli, the accuracy decreased by less than 7%, and the information transfer rate increased by 60%. This research advances methods for tactile BCI control based on event-related potentials. This work is significant since tactile stimulation is an understudied modality for BCI control, and electrically induced sERPs are the least studied control signals in reactive BCIs. Exploring and optimizing the parameters of sERP elicitation, as well as feature extraction and classification methods, is crucial for addressing the accuracy versus speed trade-off in various assistive BCI applications where the tactile modality may have added value.

EEG channel and feature investigation in binary and multiple motor imagery task predictions

IntroductionMotor Imagery (MI) Electroencephalography (EEG) signals are non-stationary and dynamic physiological signals which have low signal-to-noise ratio. Hence, it is difficult to achieve high classification accuracy. Although various machine learning methods have already proven useful to that effect, the use of many features and ineffective EEG channels often leads to a complex structure of classifier algorithms. State-of-the-art studies were interested in improving classification performance with complex feature extraction and classification methods by neglecting detailed EEG channel and feature investigation in predicting MI tasks from EEGs. Here, we investigate the effects of the statistically significant feature selection method on four different feature domains (time-domain, frequency-domain, time-frequency domain, and non-linear domain) and their two different combinations to reduce the number of features and classify MI-EEG features by comparing low-dimensional matrices with well-known machine learning algorithms.MethodsOur main goal is not to find the best classifier performance but to perform feature and channel investigation in MI task classification. Therefore, the detailed investigation of the effect of EEG channels and features is implemented using a statistically significant feature distribution on 22 EEG channels for each feature set separately. We used the BCI Competition IV Dataset IIa and 288 samples per person. A total of 1,364 MI-EEG features were analyzed in this study. We tested nine distinct classifiers: Decision tree, Discriminant analysis, Logistic regression, Naive Bayes, Support vector machine, k-Nearest neighbor, Ensemble learning, Neural networks, and Kernel approximation.ResultsAmong all feature sets considered, classifications performed with non-linear and combined feature sets resulted in a maximum accuracy of 63.04% and 47.36% for binary and multiple MI task predictions, respectively. The ensemble learning classifier achieved the maximum accuracy in almost all feature sets for binary and multiple MI task classifications.DiscussionOur research thus shows that the statistically significant feature-based feature selection method significantly improves the classification performance with fewer features in almost all feature sets, enabling detailed and effective EEG channel and feature investigation.