Abstract COVID‐19 pandemic has impacted many nations, causing physical as well as mental health concerns globally. In most countries, governments enforced strict lockdowns and social distancing, thus affecting people's daily lives. People usually tweet their views on online platforms that is unstructured text with implicit meaning. With the evolution of artificial intelligence in the natural language processing domain, the prediction of sentiments accurately has become a challenge. To contribute as a solution to this, a hybrid approach is proposed for sentiment prediction with the use of an evolutionary‐based approach, transfer‐based learning and machine learning. The proposed approach uses bidirectional encoder representations from transformers (BERT) with genetic algorithm (GA) and support vector machine (SVM), namely, hybrid evolutionary intelligent model (GA‐BERT‐SVM). These approaches aid in extracting important features considering semantics and context present in the text. To avoid the limitations of the backpropagation approach, such as trapping in local minima and overfitting the data, the initial parameters (weights and biases) of the dense layers has been optimized using GA. Additionally, the pretrained BERT layers are utilized without any modification, following a standard transfer learning approach. The BERT embeddings are concatenated with the SVM for training and classification. GridSearchCV and GeneticSearchCV is used for obtaining optimal parameters of SVM. A multi‐classification problem is tackled using a benchmark COVID‐19 dataset, which comprises of Twitter data and is categorized into COVIDSENTI‐A, COVIDSENTI‐B, COVIDSENTI‐C and a combined dataset called COVIDSENTI. Experimental evaluation demonstrates promising results of the proposed model in terms of accuracy, F1‐score, precision and recall, surpassing state‐of‐the‐art approaches.

Abstract Mesothelioma is an aggressive lung cancer, harms the linings of the lungs. It is one of the deadliest cancers diagnosed in those exposed to fibrous silicate minerals (asbestos). Millions of people face severe consequences as they are diagnosed at late stages. This study presents a comparison of several machine learning approaches with distinct feature sets and addresses the issue of class imbalance. The dataset used in this study is available publicly on the University of California Irvine (UCI) machine learning repository. This study used the resampling technique, synthetic minority oversampling technique (SMOTE), and adaptive synthetic sampling (ADASYN) to handle the class imbalance. Most of the machine learning strategies performed well with the resampling technique. The best accuracy using the resampling strategy was achieved by artificial neural networks (ANN). The highest accuracy was recorded on the feature set selected by principal component analysis (PCA) is 96%. Overall, ensemble techniques performed well. The proposed stacking-based classifier achieved the highest accuracy (89%) on data balanced using SMOTE and ADASYN.

One of the most prevalent mental disorders, i.e., depression, affects millions of individuals worldwide. Most people who are depressed are hesitant to reveal their condition to others. The main objective of this study is to predict the depression of online users using Natural Language Processing (NLP) tools and deep learning techniques. This paper proposes a approach focusing sentiment classification which is a hybrid of Long Short-Term Memory (LSTM) and Temporal Convolution Network (TCN) models. The proposed LSTM-TCN model addresses the limitations of traditional Convolutional Neural Networks (CNNs), i.e., the inability of CNNs to fully capture text features during feature extraction. Furthermore, a single model cannot properly extract deep text features. A publicly available large-scale dataset comprising Twitter data is utilized for experimentation purposes. A comparison analysis is performed illustrating that the proposed LSTM-TCN model outperforms traditional neural networks namely, CNN, RNN, Enhanced RCNN and DNN-LSTM with an accuracy of 96.96%.

Abstract Cancer is one of the leading causes of death in the world. Cancer research is vital as the prognosis of cancer enables clinical applications for patients. In this study, we have proposed the Stacked Ensemble Model (Stacking of bagged and boosted learners) for the automatic disease diagnosis. The experimental results prove the superiority of the proposed method to conventional machine learning techniques. In the empirical study, the performance of eight data handling methods and 14 classification methods is compared to obtain prediction results. The performance of the model has been evaluated on five benchmark datasets. The appreciable Area under the Curve scores achieved by the proposed methodology on Cervical Cancer (0.98), Mesothelioma (0.93), Breast Cancer (0.99), Prostate Cancer (0.97), and Hepatitis‐C Virus (0.998) datasets make this work more significant than the previously published works. The experimental results show that our proposed method is superior to conventional machine learning techniques and the proposed model contributes in the form of an efficient computational model.

Abstract Cervical cancer is one of the most common cancers among women in the world. As at the earlier stage, cervical cancer has fewer symptoms. Cancer research is vital as the prognosis of cancer enables clinical applications for patients. In this study, we demonstrate a new approach that applies an ensemble approach to machine learning models for the automatic diagnosis of cervical cancer. The dataset used in the study is the cervical cancer dataset available at the University of California Irvine database repository. Initially, missing values are imputed (k-nearest neighbors) and then the data are balanced (oversampled). Two feature selection approaches are used to extract the most significant features. The proposed stacking architecture, applied for the first time on the cervical cancer dataset, used time elapse of 5.6 s and achieved an area under the curve score of 99.7% performing better than the methods used in previous works. The objective of the study is to propose a computational model that can predict the diagnosis of cervical cancer efficiently. Further, the proposed learning architecture is gauged with several ensemble approaches like random forest, gradient boosting, voting ensemble and weighted voting ensemble to perceive the enhancement.

Abstract A class imbalance problem plays a vital role while dealing with classes with rare number of instances. Noisy class imbalanced datasets create considerable effect on the machine learning classification of classes. Data resampling techniques commonly used for handling class imbalance problem show insignificant behavior in noisy imbalanced datasets. To cure curse of data resampling technique in noisy class imbalanced data, we have proposed improved hybrid bag-boost with proposed resampling technique model. This model contains proposed resampling technique used for handling noisy imbalanced datasets. Proposed resampling technique comprises K-Means SMOTE (Synthetic Minority Oversampling TEchnique) as an oversampling technique and edited nearest neighbor (ENN) undersampling technique used as noise removal. This resampling technique is used to mitigate noise in imbalanced datasets at three levels, i.e. first clusters datasets using K-Means clustering technique, SMOTE inside clusters for handling imbalance by inducing synthetic instances of class in minority and lastly, using ENN technique to remove instances that create noise afterwards. Experiments were performed using 11 binary imbalanced datasets by varying attribute noise percentages, and by using area under receiver operating curve as performance metrics. Experimental results confirmed that proposed model shows better results than the rest. Moreover, it is also confirmed that proposed technique performs better with an increased noise percentage in binary imbalanced datasets.

SUMMARY Depression is a clinical entity that might be difficult for a psychiatrist to diagnose it effectively on time. A depressed person usually suffers from distress and anxiety, leading to serious consequences if not diagnosed early. Social media platforms facilitate users to exchange ideas and dialogs, resulting in the collection of a huge volume of data. Analyzing user's online behavior to categorize depression is a challenging task for researchers. This motivated researchers to investigate machine learning, deep learning, and natural language processing techniques supporting research related to depression prediction. The dataset used in the study is a large‐scale Twitter dataset. This article aims to investigate a hybrid CNN‐LSTM deep learning model with the Word2Vec feature extraction technique for classifying depressive sentiments from Twitter data. By using TF‐IDF, PCA, and Word2Vec approaches, this model utilizes significant linguistic features present within the text. The proposed model is evaluated on four benchmark datasets and its efficiency is compared with four traditional machine learning models. Moreover, the proposed model's performance is compared to three deep learning‐based hybrid models. The proposed model showed comparable performance with the hybrid deep learning‐based models and outperformed state‐of‐the‐art machine learning techniques with an accuracy of 96.78% and an MSE score of 3.21.

Cancer is one of the most significant causes of death worldwide, accounting for millions of deaths each year. The fatality rate of cancer is getting higher. Over the last three decades, deep neural networks have been critical in cancer research. This article described the development of a system for fully automated segmentation of brain tumor. In this study, we have proposed a unique ensemble of Convolutional Neural Networks (ConvNet) for segmenting gliomas from MR images. Two fully linked ConvNets constituted the ensemble model (2D-ConvNet and 3-D ConvNet). The novel model is validated against a single dataset from the Brain Tumor Segmentation (BraTS) challenge, specifically BraTS_2018. The prediction results obtained using the proposed methodology on the BraTS_2018 datasets demonstrate the suggested architecture's efficiency.