Search

The integration of the Internet of Things in medical systems referred to as the Internet of Medical Things (IoMT), which supports medical events for instance real-time diagnosis, remote monitoring of patients, real-time drug prescriptions, among others. This aids the quality of services provided by the health workers thereby improve patients’ satisfaction. However, the integrity and confidentiality of medical information on the IoMT platform remain one of the contentions that causes problems in medical services. Another serious concern with achieving protection for medical records is information confidentiality for patient’s records over the IoMT environment. Therefore, this paper proposed a Crypto-Stegno model to secure medical information on the IoMT environment. The paper validates the system on healthcare information datasets and revealed extraordinary results in respect to the quality of perceptibility, extreme opposition to data loss, extreme embedding capability and security, which made the proposed system an authentic strategy for resourceful and efficient medical information on IoTM platform.

PurposeHate speech is an expression of intense hatred. Twitter has become a popular analytical tool for the prediction and monitoring of abusive behaviors. Hate speech detection with social media data has witnessed special research attention in recent studies, hence, the need to design a generic metadata architecture and efficient feature extraction technique to enhance hate speech detection.Design/methodology/approachThis study proposes a hybrid embeddings enhanced with a topic inference method and an improved cuckoo search neural network for hate speech detection in Twitter data. The proposed method uses a hybrid embeddings technique that includes Term Frequency-Inverse Document Frequency (TF-IDF) for word-level feature extraction and Long Short Term Memory (LSTM) which is a variant of recurrent neural networks architecture for sentence-level feature extraction. The extracted features from the hybrid embeddings then serve as input into the improved cuckoo search neural network for the prediction of a tweet as hate speech, offensive language or neither.FindingsThe proposed method showed better results when tested on the collected Twitter datasets compared to other related methods. In order to validate the performances of the proposed method, t-test and post hoc multiple comparisons were used to compare the significance and means of the proposed method with other related methods for hate speech detection. Furthermore, Paired Sample t-Test was also conducted to validate the performances of the proposed method with other related methods.Research limitations/implicationsFinally, the evaluation results showed that the proposed method outperforms other related methods with mean F1-score of 91.3.Originality/valueThe main novelty of this study is the use of an automatic topic spotting measure based on naïve Bayes model to improve features representation.

This paper employed the unified theory of acceptance and use of technology (UTAUT) method to inspect integrating SM and exploit for learning purposes among university students in Nigeria. The test result showed that social and peer influence are pertinent elements that have an impact on social media (SM) recognition for education intentions. Consequently, the parent or institutional management can refuse students access to SM, if employed for erroneous intentions instead of acquiring purposes. Furthermore, the outcome showed that each of the five independent variables has a straight consequence on the education aim employed. Similarly, the findings equally showed that SM environments possess a clear, postulated impact on education use. This can be a reason that not all SMs can be conveniently utilized for the intention of learning. Furthermore, student experience could as well be influenced whenever SM was viewed exclusively for social undertakings which may include viewing movies, participating in games, and being linked to relatives as well as acquaintances. Also, the current study expands the original UTAUT by evaluating the influence of two additional variables (peer pressure and conditions of social media). Eventually, the study conceived a general questionnaire to assess factors that incorporate SM and use them for learning purposes.

Computing technologies perform a vital role in the transformation of contemporary financial services. The emergence of financial technology (fintech) in recent years has transformed different facets of the global financial industries, thus changing the way financial personnel lives, think, and work. The fintech similarly had developed numerous technologies to advance the industry. These developments focused on the advancement of financial segments through the use of big data, cloud computing, the Internet of Things, artificial intelligence, and blockchain, for better transaction and security. Big data has seen a substantial adjustment in approaches to gather, demonstrate, and appraise information. How to use the data to develop new business models has attracted considerable attention, and recent technological advances have allowed us to produce data much faster than ever. Big data analytics is the modeling of big data to discover useful information and patterns, thereby communicating the same to support decision making and suggest conclusions. Big data can be processed, accumulated, and accumulated in the financial industries through rapidly evolving fields of big data analytics, thus enabling the understanding, rationalization, and use of big data for different purposes. Therefore, this chapter discusses the concept of big data for guaranteeing better expansion and research toward fintech. The chapter likewise will examine how large-volume financial rights data can greatly help clarify and elucidate fintech patterns. The chapter will also make an effort to convey various benefits that financial industries can make by using fintech with big data technology. The hope of using fintech with big data in financial services will have a great impact on the quality of financial transactions that can be delivered to financial sectors across socioeconomic and geographic boundaries.

The IoT revolution reshapes contemporary healthcare systems by incorporate economic, social, and technological prospects. It is progressing from conventional healthcare systems to more personalized healthcare systems, where patients can be monitored, diagnosed and treated effortlessly. Radiomics is a sub-field of machine learning (ML) that mines quantitative features from radiological images relying on an image-based approach with ML models, which procure information surpassing orthodox medical imaging analysis as diagnosis, prognosis, prediction and response to therapy. The upsurge in the number of radiological images increases the workload of radiologist which in turns decreases their performance, thus they can only detect and evaluate a small portion of information present in images within a short-time. Hence, there is need for a better method for the increase in radiological image selection, detection and evaluation processes thereby reducing the workload of experts. Therefore, this chapter discusses the different types and sources of radiological data, feature extraction and selection method for image analysis. The chapter also presents different ML models ideal for the radiomics and parameter tuning. The challenges, applicability and limitations of Radiomics are also described in this chapter. The radiomic process involves radiological image gathering, segmentation, feature extraction and selection, model building and evaluation. Each of the stage of the process workflow is carefully evaluated for development of a reliable, effective and robust model to be shifted into medical practice for disease diagnosis and prognosis response to treatment.

Diabetes mellitus (DM) is one of the chronic and debilitating diseases in modern society, hence the urgent need to prevent epidemic growth in society. This chapter is motivated by the studies of several scholars in the field of microarrays datasets for gene expression. Nonetheless, there are very few available gene signatures across datasets, thereby generating sample selection bias and over selection sets matching. Subjective selection of this gene and sample pairings could be addressed through large average submatrices and a unique method of biclustering using objective statistical assumptions to reconstruct robust signatures of expression. Hence, SWITCH (SupWald Identification of CHanges DNA copy) was created to label CNAs in platforms of aCGH and to connect them with subtypes. Therefore, the process of selecting the most informative gene biomarker was done using the genetic algorithm (GA) and deep neural networks (DNN) for biological sample classification. The simulated genomics datasets were divided into 95% training and 5% test samples and the DNN classifier is modified using these sets of SNPs and fine-tuned to classify type II DM analyses. The datasets are cleaned into four single-nucleotide polymorphism (SNP) function sets: 96 (P-value: 1 × 10− 5), 214 (P-value: 1 × 10− 4), 399 (P-value: 1 × 10− 3), and 678 (P-value: 1 × 10− 2) using P-value thresholds. The classifier was built using the training data while testing its efficiency on the test sample. MATLAB was used to implement the GA and DNN. The DNN model showed a significant predictive output with type II DM having AUC = 0.9537 in male and AUC = 0.9349 in female. An experimental test was carried out to determine the associations of all SNP datasets with the type II diabetes phenotype. The results from the test showed an enhanced model performance with 399 and 678 SNPs, respectively. The test result also showed that the higher the number of SNPs the better the predictive performance of the designed model. Finally, the result showed that DNNs can be used to predict type II diabetes using genomic-based data.

PurposeOver the past decade, the cost of product development has increased drastically, and this is due to the inability of most enterprises to locate suitable and optimal collaborators for knowledge sharing. Nevertheless, knowledge sharing is a mechanism that helps people find the best collaborators with relevant knowledge. Hence, a new approach for locating optimal collaborators with relevant knowledge is needed, which could help enterprise in reducing cost and time in a knowledge-sharing environment. The paper aims to discuss these issues.Design/methodology/approachOne unique challenge in the domain of knowledge sharing is that collaborators do not possess the same number of events resident in the knowledge available for sharing. In this paper, the authors present a new approach for locating optimal collaborators in knowledge-sharing environment using the combinatorial algorithm (CA-KSE).FindingsThe proposed pattern-matching approach implemented in Java is considered efficient for solving the issue peculiar to collaboration in knowledge-sharing domain. The authors benchmarked the proposed approach with its semi-global pairwise alignment and global alignment counterparts through scores comparison and the receiver operating characteristic curve. The results obtained from the comparisons showed that CA-KSE is a perfect test having an area under curve of 0.9659, compared to the other approaches.Research limitations/implicationsThe paper has proposed an efficient algorithm, which is considered better than related methods, for matching several collaborators (more than two) in KS environment. The method could be deployed in medical field for gene analysis, software organizations for distributed development and academics for knowledge sharing.Originality/valueOne sign of strength of this approach, compared to most sequence alignment approaches that can only match two collaborators at a time, is that it can match several collaborators at a faster rate.