Correlation Feature Selection Research Articles

Application of Stacking Ensemble Learning in Predicting Copper’s Flotation Concentrate Grade

Addressing issues such as a low operational condition recognition efficiency, strong subjectivity, and significant fluctuations in Outotec X fluorescence analysis data in copper flotation production, a copper concentrate grade classification model is constructed based on image processing technology and the Stacking ensemble learning algorithm. Firstly, a feature extraction model for copper concentration flotation foam images is established, extracting color, texture, and size statistical features to build a feature dataset. Secondly, to avoid redundancy in the feature data, which could reduce model accuracy, a combined correlation feature selection is employed for dimensionality reduction, with the filtered feature subset being used as the model input. Finally, to fully leverage the strengths of each model, a Stacking ensemble learning copper concentrate grade classification model is constructed with support vector machine (SVM), random forest (RF), and adaptive boosting (AdaBoost) as base models and logistic regression (LR) as the meta-model. The experimental results show that this ensemble model achieves good recognition for different grade categories, with a precision, recall, and F1 score of 90.01%, 89.85%, and 89.93%, respectively. The accuracy of this Stacking ensemble model, with a 7% improvement over Outotec X fluorescence analysis, demonstrates a potential to meet the daily production needs of beneficiation plants.

Optimizing bearing health condition monitoring: exploring correlation feature selection algorithm

Vibration signals are a critical source of information for detecting and diagnosing bearing faults, making this research particularly relevant to the condition monitoring of industrial machinery, particularly bearings using vibration signals. This study delves into how feature selection can be done using Pearson’s Correlation Co-efficient within the context of monitoring bearing health conditions, utilizing two distinct approaches. Approach-1 involves feature selection without considering labels, while Approach-2 incorporates labels for feature selection. Comparative analysis is conducted against outcomes obtained when all features are selected. The research scrutinizes the impact of feature selection on classifier performance, accuracy, and execution times, utilizing various machine learning algorithms such as Decision Tree (DT), K Nearest Neighbor (KNN), Support Vector Machine (SVM), and Naïve Bayes (NB). The findings underscore that feature selection significantly enhances classifier accuracy while reducing execution times. Specifically, only DT and KNN with 50 neighbors achieved 100% accuracy when all features were considered. However, with feature selection using Approach-1 (without labels), DT, KNN, SVM (excluding 100 neighbors), and NB (with Normal/Gaussian kernel) attained 100% accuracy. Employing Approach-2 (with labeled features), DT with 0.7 and 0.9 thresholds, SVM-G with all thresholds (0.6, 0.7, and 0.9), KNN with all thresholds (except 100 neighbors), and NB-n (with all thresholds) achieved 100% accuracy. The study emphasizes the pivotal role of feature selection using Pearson’s Correlation Coefficient in enhancing machine learning classifier performance, offering promising avenues for future research and practical applications across diverse domains.

Developing a hybrid feature selection method to detect botnet attacks in IoT devices

The Internet of Things, or IoT, is an important technology applied in various applications such as smart homes and innovative healthcare. Due to its architecture, IoT-based devices suffer from various security challenges, most commonly, botnet attacks. This article aims to develop a hybrid feature selection method to find the most influential features based on three feature selection methods, correlation, generalized normal distribution optimization, and lasso, to detect botnet attacks in IoT devices. The UNSW-NB15 dataset is used to assess the proposed system. Several classification models including decision tree (DT), random forest (RF), k-nearest neighbors (KNN), adaptive boosting (AdaBoost), and bagging are utilized for the classification purpose. The proposed system was evaluated using several performance metrics. The results showed the correlation feature selection method had the most accurate botnet attack detection rate. RF also outperformed other models with a 95.11% detection rate in binary classification and 83.96% in multi-classification. On the other hand, results showed that the proposed hybrid method outperformed the feature selection methods with an increase of about 3% in both classifications. The AdaBoost model achieved an accuracy of 99.28% with binary classification by using 18 features, and the RF model achieved an accuracy of 86.62% with multi-classification by using 22 features. The robustness and efficacy of the proposed approach were demonstrated by comparing the study's results with several other studies that have used the same dataset. The results of the study can be implemented in real applications to detect network interference of a dynamic nature in real-time and assist intrusion detection systems (IDS) in addressing these attacks.

Non-convex feature selection based on feature correlation representation and dual manifold optimization

The existing feature selection algorithms often utilize local structure of data, but do not fully mine internal structure and ignore potential correlation information between samples. To address the problems and fully utilize manifold information of data, this paper proposes non-convex feature selection based on feature correlation representation and dual manifold optimization (FDNFS). Firstly, FDNFS constructs feature graph of original data, which can obtain feature correlation representation to represent the interconnection information. Based on the obtained interconnection information, FDNFS unifies feature correlation representation learning and feature selection through feature transformation matrix, so that the interconnection information between data guides feature selection. Secondly, to make multivariate frameworks guide feature selection, FDNFS introduces self-representation on the improved sparse regression model. Using self-representation can make basis matrix and reconstruction coefficient matrix reconstruct original data more accurately. Next, in order to preserve local structure information abundantly, FDNFS has two-part manifold regularization on the pseudo-label matrix in sparse regression model and the reconstructed coefficient matrix in self-representation framework. This can fully use the manifold information of data. In addition, FDNFS imposes the non-convex constraints. It can ensure the sparsity of feature selection matrix. In turn, this can select features with lower redundancy, and then select a better feature subset. Finally, this paper adopts an iterative optimization method. FDNFS is compared with nine algorithms on seven datasets. The clustering results reflect better performance of FDNFS.

Statistical methods of feature engineering for the problem of forest state classification using satellite data

Timely detection of forest diseases is an important task for their prevention and spread limitation. The usage of satellite imagery provides capabilities for large-scale forest monitoring. Machine learning models allow to automate the analysis of these data for anomaly detection indicating diseases. However, selecting informative features is key to building an effective model. In this work, the application of Bhattacharyya distance and Spearman’s rank correlation coefficient for feature selection from satellite images was investigated. A greedy algorithm was applied to form a subset of weakly correlated features. The experiment showed that selected features allow for improving the classification quality compared to using all spectral bands. The proposed approach demonstrates effectiveness for informative and weakly correlated feature selection and can be utilized in other remote sensing tasks.

Classification of breast cancer data based on neural network algorithm

In today's world, breast cancer has become the most common malignant tumor, so this paper mainly classifies the features of breast cancer through BP neural network to improve the accuracy of research and judgment. This paper mainly adopts 30 breast cancer feature data of 569 cases, and then obtains the feature data used in this BP neural network through correlation analysis, feature selection and principal component analysis. By training, the BP neural network that is most suitable for this data is constructed from aspects such as the number of hidden layers, loss function, and iteration times, ultimately improving the accuracy to 100%.

Abstract 4930: Identification of CBX7 and PCDHB18 as novel prognostic biomarkers of cervical cancer: RNA-sequencing and machine learning analysis

Abstract Background- Cervical cancer, ranking fourth in prevalence among women worldwide, is highly preventable when diagnosed early. The advent of innovative technologies including bioinformatics and machine learning, is revolutionizing the discovery and development of novel biomarkers for cancer. Our study uniquely integrates artificial neural networks with RNA-sequencing data of cervical cancer patients. This combination enhances the accuracy and reliability of biomarker predictions, providing a better understanding of the potential clinical utility of the identified biomarkers. Methods- RNA-sequencing data and clinicopathological details of 304 Cervical Squamous Cell Carcinoma and Endocervical Adenocarcinoma (CESC) were obtained from the GDAC database (https://gdac.broadinstitute.org/). Differentially expressed genes (DEGs) were identified (P<0.05, |log2 fold change (FC)| > 1.5, false discovery rate (FDR) < 0.05). Pathway enrichment analysis was performed and protein-protein interactions of DEGs were constructed using the STRING database. Prognostic biomarkers were identified using Kaplan-Meier and Cox proportional hazard methods adjusted for potential confounders such as age, disease stage, and comorbidities (HR<1, p<0.05). Deep learning algorithms were applied for predictive marker identification, utilizing Weight by Correlation feature selection. The model used AUC (Area Under the Curve), accuracy, MSE (Mean Squared Error), and R2 (R-squared) as evaluation metrics in a 70-30 training-test split. A combined ROC curve assessed diagnostic biomarkers, validated externally with a GEO dataset on cervical cancer patients. Results- In our study, 4153 DEGs were identified. Pathway analysis revealed that key dysregulated genes play a pivotal role in extracellular matrix organization. The survival analysis identified that six upregulated genes (CT62, SLC7A5P1, C10orf110, FDPSL2A, TNFSF15, and TTLL13) and seven downregulated genes (BAI3, CBX7, DKFZp566F0947, PCDHB18, GRAPL, PCDHB19P, and C4orf38) decreased the overall survival in patients. The machine learning model demonstrated high predictive accuracy (AUC=1, accuracy=99.02%, R2=0.99), identifying twenty genes with a positive correlation to cervical cancer risk. Notably, CBX7 emerged as a prognostic and diagnostic biomarker (AUC=0.99, sensitivity=0.93, specificity=1.00). Conclusion- Our study uncovers the prognostic significance of two novel genes in cervical cancer: CBX7, a vital regulator of tumor suppression, and PCDHB18, a member of the protocadherin beta gene cluster functioning as a cell adhesion molecule. Downregulation of these genes is associated with decreased overall survival. Further functional analyses and validation of these candidate biomarkers are crucial to fully assess their potential clinical value in cervical cancer. Citation Format: Ghazaleh Pourali, Mohsen Zeinali, Mahshid Arastonejad, Nima Khalili-Tanha, Elham Nazari, Ghazaleh Khalili-Tanha, Adetunji T. Toriola. Identification of CBX7 and PCDHB18 as novel prognostic biomarkers of cervical cancer: RNA-sequencing and machine learning analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4930.

Transformer Core Fault Diagnosis via Current Signal Analysis with Pearson Correlation Feature Selection

The role of transformers in power distribution is crucial, as their reliable operation is essential for maintaining the electrical grid’s stability. Single-phase transformers are highly versatile, making them suitable for various applications requiring precise voltage control and isolation. In this study, we investigated the fault diagnosis of a 1 kVA single-phase transformer core subjected to induced faults. Our diagnostic approach involved using a combination of advanced signal processing techniques, such as the fast Fourier transform (FFT) and Hilbert transform (HT), to analyze the current signals. Our analysis aimed to differentiate and characterize the unique signatures associated with each fault type, utilizing statistical feature selection based on the Pearson correlation and a machine learning classifier. Our results showed significant improvements in all metrics for the classifier models, particularly the k-nearest neighbor (KNN) algorithm, with 83.89% accuracy and a computational cost of 0.2963 s. For future studies, our focus will be on using deep learning models to improve the effectiveness of the proposed method.

An empirical analysis of feature selection techniques for Software Defect Prediction

<p>Detecting software defects before they occur is crucial in software engineering as it impacts software system quality and reliability. Previous studies on predicting software defects have typically employed software features, such as code size, complexity, coupling, cohesion, inheritance, and other software metrics., to forecast whether a code file or commit is prone to defects in the future. However, it is advantageous to restrict the number of features employed in a defect prediction model to avoid the challenges associated with multicollinearity and the “curse of dimensionality” and to simplify the data analysis process. By using a reduced number of features, the defect prediction model can concentrate on the most significant variables and improve its accuracy. This research paper investigates the impact of eight feature selection methods on the accuracy and stability of six supervised learning models. This study is novel as it is based on exhaustive experimentation of each of the eight feature selection techniques with each of the six supervised learning models. Two notable findings have been obtained. First, we discovered that the association and coherence-based techniques have demonstrated the highest level of accuracy when it comes to defect prediction. The models that utilized these selected features outperformed those using the original features. Second, the feature selection techniques, namely Correlation feature selection, Recursive feature elimination, and Ridge feature selection when combined with the Support vector machine and Decision tree classifier, consistently selected low-variance features across multiple supervised defect prediction models. When combined with different classifiers, these techniques achieved exceptional performance on the publicly available NASA datasets CM1 and PC2. The findings revealed a remarkable accuracy rate of over 85% for CM1 and 95% for PC2, accompanied by precision, recall, and f-measure values exceeding 95%. These exceptional results indicate the achievement of the highest level of performance in the evaluation.</p>

Innovative Teaching of College Vocal Music Courses Aided by Information Integration Technology

Abstract The article adopts the correlation function method to extract the fundamental frequency features of vocal speech signals and processes this information through the cepstrum method. Then, the audio signals are corrected by linear smoothing to match the features of the students’ audio signals with the features of the musical score to determine the accuracy of the vocal music. The study also includes quantization, frame-splitting, and windowing preprocessing of the audio signals, and constructing an unwritten audio signal feature set based on correlation feature selection. The article also proposes an innovative metaphorical method for improving the teaching mode of vocal music courses in colleges and universities. It analyzes the auxiliary role of information technology in teaching vocal music courses by combining simulation experiments with the effects of practical application. The study results show that the students’ vocal singing fundamental frequency range is close to the standard, roughly 123 to 200, with a maximum error value of 24. With the assistance of information fusion technology, the average score increment of the students’ vocal singing reaches 6.088 points, proving this method’s effectiveness in improving the quality of verbal music teaching.

An efficient feature selection and explainable classification method for EEG-based epileptic seizure detection

Epilepsy is a prevalent neurological disorder that poses life-threatening emergencies. Early electroencephalogram (EEG) seizure detection can mitigate the risks and aid in the treatment of patients with epilepsy. EEG based epileptic seizure (ES) detection has significant applications in epilepsy treatment and medical diagnosis. Therefore, this paper presents an innovative framework for efficient ES detection, providing coefficient and distance correlation feature selection algorithms, a Bagged Tree-based classifer (BTBC), and Explainable Artificial Intelligence (XAI). Initially, the Butterworth filter is employed to eliminate various artifacts, and the discrete wavelet transform (DWT) is used to decompose the EEG signals and extract various eigenvalue features of the statistical time domain (STD) as linear and Fractal dimension-based non-linear (FD-NL). The optimal features are then identified through correlation coefficients with P-value and distance correlation analysis.These features are subsequently utilized by the Bagged Tree-based classifer (BTBC). The proposed model provides best performance in mitigating overfitting issues and improves the average accuracy by 2% using (CD, E), (AB, CD, E), and (A, B) experimental types as compared to other machine learning (ML) models using well-known Bonn and UCI-EEG benchmark datasets. Finally, SHapley additive exPlanation (SHAP) was used to interpret and explain the decision-making process of the proposed model. The results highlight the framework’s capability to accurately classify ES, thereby improving the diagnosis process in patients with brain dysfunctions.

Temporal Convolutional Networks and BERT-Based Multi-Label Emotion Analysis for Financial Forecasting

The use of deep learning in conjunction with models that extract emotion-related information from texts to predict financial time series is based on the assumption that what is said about a stock is correlated with the way that stock fluctuates. Given the above, in this work, a multivariate forecasting methodology incorporating temporal convolutional networks in combination with a BERT-based multi-label emotion classification procedure and correlation feature selection is proposed. The results from an extensive set of experiments, which included predictions of three different time frames and various multivariate ensemble schemes that capture 28 different types of emotion-relative information, are presented. It is shown that the proposed methodology exhibits universal predominance regarding aggregate performance over six different metrics, outperforming all the compared schemes, including a multitude of individual and ensemble methods, both in terms of aggregate average scores and Friedman rankings. Moreover, the results strongly indicate that the use of emotion-related features has beneficial effects on the derived forecasts.

Stock price indices prediction combining deep learning algorithms and selected technical indicators based on correlation

PurposeThe prediction of stock market (SM) indices is a fascinating task. An in-depth analysis in this field can provide valuable information to investors, traders and policy makers in attractive SMs. This article aims to apply a correlation feature selection model to identify important technical indicators (TIs), which are combined with multiple deep learning (DL) algorithms for forecasting SM indices.Design/methodology/approachThe methodology involves using a correlation feature selection model to select the most relevant features. These features are then used to predict the fluctuations of six markets using various DL algorithms, and the results are compared with predictions made using all features by using a range of performance measures.FindingsThe experimental results show that the combination of TIs selected through correlation and Artificial Neural Network (ANN) provides good results in the MADEX market. The combination of selected indicators and Convolutional Neural Network (CNN) in the NASDAQ 100 market outperforms all other combinations of variables and models. In other markets, the combination of all variables with ANN provides the best results.Originality/valueThis article makes several significant contributions, including the use of a correlation feature selection model to select pertinent variables, comparison between multiple DL algorithms (ANN, CNN and Long-Short-Term Memory (LSTM)), combining selected variables with algorithms to improve predictions, evaluation of the suggested model on six datasets (MASI, MADEX, FTSE 100, SP500, NASDAQ 100 and EGX 30) and application of various performance measures (Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error(RMSE), Mean Squared Logarithmic Error (MSLE) and Root Mean Squared Logarithmic Error (RMSLE)).

Circular-SWAT for deep learning based diagnostic classification of Alzheimer's disease: application to metabolome data

Deep learning has shown potential in various scientific domains but faces challenges when applied to complex, high-dimensional multi-omics data. Alzheimer's Disease (AD) is a neurodegenerative disorder that lacks targeted therapeutic options. This study introduces the Circular-Sliding Window Association Test (c-SWAT) to improve the classification accuracy in predicting AD using serum-based metabolomics data, specifically lipidomics. The c-SWAT methodology builds upon the existing Sliding Window Association Test (SWAT) and utilizes a three-step approach: feature correlation analysis, feature selection, and classification. Data from 997 participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI) served as the basis for model training and validation. Feature correlations were analyzed using Weighted Gene Co-expression Network Analysis (WGCNA), and Convolutional Neural Networks (CNN) were employed for feature selection. Random Forest was used for the final classification. The application of c-SWAT resulted in a classification accuracy of up to 80.8% and an AUC of 0.808 for distinguishing AD from cognitively normal older adults. This marks a 9.4% improvement in accuracy and a 0.169 increase in AUC compared to methods without c-SWAT. These results were statistically significant, with a p-value of 1.04×10ˆ-4. The approach also identified key lipids associated with AD, such as Cer(d16:1/22:0) and PI(37:6). Our results indicate that c-SWAT is effective in improving classification accuracy and in identifying potential lipid biomarkers for AD. These identified lipids offer new avenues for understanding AD and warrant further investigation. The specific funding of this article is provided in the acknowledgements section.

Enhancing Sustainable Urban Energy Management through Short-Term Wind Power Forecasting Using LSTM Neural Network

Integrating wind energy forecasting into urban city energy management systems offers significant potential for optimizing energy usage, reducing the carbon footprint, and improving overall energy efficiency. This article focuses on developing a wind power forecasting model using cutting-edge technologies to enhance urban city energy management systems. To effectively manage wind energy availability, a strategy is proposed to curtail energy consumption during periods of low wind energy availability and boost consumption during periods of high wind energy availability. For this purpose, an LSTM-based model is employed to forecast short-term wind power, leveraging a publicly available dataset. The LSTM model is trained with 27,310 instances and 10 wind energy system attributes, which were selected using the Pearson correlation feature selection method to identify crucial features. The evaluation of the LSTM-based forecasting model yields an impressive R2 score of 0.9107. The model’s performance metrics attest to its high accuracy, explaining a substantial proportion of the variance in the test data. This study not only contributes to advancing wind power forecasting, but also holds promise for sustainable urban energy management, enabling cities to make informed decisions in optimizing energy consumption and promoting a greener, more resilient future.

Towards data security assessments using an IDS security model for cyber-physical smart cities

Technology has enabled many devices to exchange huge amounts of data and communicate with each other as Edge Intelligence in Smart Cities (EISC), as a result of rapid technological advancements. When dealing with personal data, it is paramount to ensure that it is not disclosed and that there is no disclosure of any confidential information. In recent decades, academics and industry have spent considerable time and energy discussing security and privacy. Other systems, known as intrusion detection systems, are required to breach firewalls, antivirus software, and other security equipment to provide complete system security in smart operation systems. There are three aspects to an intrusion detection system: the intrusion detection method, the architecture, and the intrusion response method. In this study, we combined linear correlation feature selection methods and cross-information. The database used in this article is KDD99. This paper examines applying two feature selection methods in predicting attacks in intrusion detection systems based on INTERACT and A multilayer perceptron (MLP). Since the number of records associated with each attack type differs, one of our suggestions is to continue using data balancing techniques. As a result, the number of records associated with each type of network status becomes closer together. The results in the categories can also be improved using information synthesis methods, such as majority voting.

Comparación de Métodos de Extracción de Características Espectrales y Dispersas para Clasificación de Sonidos Cardíacos

Cardiovascular diseases (CVDs) remain the leading cause of morbidity worldwide. The heart sound signal or phonocardiogram (PCG) is the most simple, low-cost, and effective tool to assist physicians in diagnosing CVDs. Advances in signal processing and machine learning have motivated the design of computer-aided systems for heart illness detection based only on the PCG. The objective of this work is to compare the effects of using spectral and sparse features for a classification scheme to detect the presence/absence of a pathological state in a heart sound signal, more specifically, sparse representations using Matching Pursuit with multiscale Gabor time-frequency dictionaries, linear prediction coding, and Mel-frequency cepstral coefficients. This work compares the performance of PCGs classification applying features as a result of averaging the samples or the features for each PCG sound event when feeding a random forest (RF) classifier. For data balancing, random under-sampling and synthetic minority oversampling (SMOTE) methods were applied. Furthermore, we compare the Correlation Feature Selection (CFS) and Information Gain (IG) for the dimensionality reduction. The findings show a SE=93.17 %, SP=84.32 % and ACC=85.9 % when joining MP+LPC+MFCC features set with an AUC=0.969 showing that these features are promising to be used in heart sounds anomaly detection schemes.

Multi-Label Feature Selection Via Adaptive Label Correlation Estimation

In multi-label learning, each instance is associated with multiple labels simultaneously. Multi-label data often have noisy, irrelevant, and redundant features of high dimensionality. Multi-label feature selection has received considerable attention as an effective means for dealing with high-dimensional multi-label data. Many multi-label feature selection methods exploit label correlations to help select features. However, finding label correlations and selecting features in existing multi-label feature selection methods are often two separate processes, the existence of noises and outliers in training data makes the label correlations exploited from label space less reliable. Therefore, the learned label correlations may mislead the feature selection process and result in the selection of less informative features. This article proposes a novel algorithm named ROAD, i.e., multi-label featuRe selectiOn via ADaptive label correlation estimation. ROAD jointly performs adaptive label correlation exploration and feature selection with alternating optimization to obtain reliable estimation of label correlations, which can more effectively reveal the intrinsic manifold structure among labels and lead to the selection of a more proper feature subset. Comprehensive experiments on several frequently used datasets validate the superiority of ROAD against the state-of-the-art multi-label feature selection algorithms.

Enhancing the prediction of IDC breast cancer staging from gene expression profiles using hybrid feature selection methods and deep learning architecture.

Prediction of the stage of cancer plays an important role in planning the course of treatment and has been largely reliant on imaging tools which do not capture molecular events that cause cancer progression. Gene-expression data-based analyses are able to identify these events, allowing RNA-sequence and microarray cancer data to be used for cancer analyses. Breast cancer is the most common cancer worldwide, and is classified into four stages - stages 1, 2, 3, and 4 [2]. While machine learning models have previously been explored to perform stage classification with limited success, multi-class stage classification has not had significant progress. There is a need for improved multi-class classification models, such as by investigating deep learning models. Gene-expression-based cancer data is characterised by the small size of available datasets, class imbalance, and high dimensionality. Class balancing methods must be applied to the dataset. Since all the genes are not necessary for stage prediction, retaining only the necessary genes can improve classification accuracy. The breast cancer samples are to be classified into 4 classes of stages 1 to 4. Invasive ductal carcinoma breast cancer samples are obtained from The Cancer Genome Atlas (TCGA) and Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) datasets and combined. Two class balancing techniques are explored, synthetic minority oversampling technique (SMOTE) and SMOTE followed by random undersampling. A hybrid feature selection pipeline is proposed, with three pipelines explored involving combinations of filter and embedded feature selection methods: Pipeline 1 - minimum-redundancy maximum-relevancy (mRMR) and correlation feature selection (CFS), Pipeline 2 - mRMR, mutual information (MI) and CFS, and Pipeline 3 - mRMR and support vector machine-recursive feature elimination (SVM-RFE). The classification is done using deep learning models, namely deep neural network, convolutional neural network, recurrent neural network, a modified deep neural network, and an AutoKeras generated model. Classification performance post class-balancing and various feature selection techniques show marked improvement over classification prior to feature selection. The best multiclass classification was found to be by a deep neural network post SMOTE and random undersampling, and feature selection using mRMR and recursive feature elimination, with a Cohen-Kappa score of 0.303 and a classification accuracy of 53.1%. For binary classification into early and late-stage cancer, the best performance is obtained by a modified deep neural network (DNN) post SMOTE and random undersampling, and feature selection using mRMR and recursive feature elimination, with an accuracy of 81.0% and a Cohen-Kappa score (CKS) of 0.280. This pipeline also showed improved multiclass classification performance on neuroblastoma cancer data, with a best area under the receiver operating characteristic (auROC) curve score of 0.872, as compared to 0.71 obtained in previous work, an improvement of 22.81%. The results and analysis reveal that feature selection techniques play a vital role in gene-expression data-based classification, and the proposed hybrid feature selection pipeline improves classification performance. Multi-class classification is possible using deep learning models, though further improvement particularly in late-stage classification is necessary and should be explored further.

Artificial intelligence for diagnosis of mild–moderate COVID-19 using haematological markers

Objective The persistent spread of SARS-CoV-2 makes diagnosis challenging because COVID-19 symptoms are hard to differentiate from those of other respiratory illnesses. The reverse transcription-polymerase chain reaction test is the current golden standard for diagnosing various respiratory diseases, including COVID-19. However, this standard diagnostic method is prone to erroneous and false negative results (10% -15%). Therefore, finding an alternative technique to validate the RT-PCR test is paramount. Artificial intelligence (AI) and machine learning (ML) applications are extensively used in medical research. Hence, this study focused on developing a decision support system using AI to diagnose mild-moderate COVID-19 from other similar diseases using demographic and clinical markers. Severe COVID-19 cases were not considered in this study since fatality rates have dropped considerably after introducing COVID-19 vaccines. Methods A custom stacked ensemble model consisting of various heterogeneous algorithms has been utilized for prediction. Four deep learning algorithms have also been tested and compared, such as one-dimensional convolutional neural networks, long short-term memory networks, deep neural networks and Residual Multi-Layer Perceptron. Five explainers, namely, Shapley Additive Values, Eli5, QLattice, Anchor and Local Interpretable Model-agnostic Explanations, have been utilized to interpret the predictions made by the classifiers. Results After using Pearson’s correlation and particle swarm optimization feature selection, the final stack obtained a maximum accuracy of 89%. The most important markers which were useful in COVID-19 diagnosis are Eosinophil, Albumin, T. Bilirubin, ALP, ALT, AST, HbA1c and TWBC. Conclusion The promising results suggest using this decision support system to diagnose COVID-19 from other similar respiratory illnesses.