Optimal deep learning approaches and healthcare big data analytics for mobile networks toward 5G

• Published in: Computers & electrical engineering Vol. 95; p. 107376
• Main Authors: Pustokhin, Denis A., Pustokhina, Irina V., Rani, Poonam, Kansal, Vineet, Elhoseny, Mohamed, Joshi, Gyanendra Prasad, Shankar, K.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier Ltd 01.10.2021; Elsevier BV
• Subjects: 5G mobile communication; 5G networking; Algorithms; Artificial intelligence; Belief networks; Big Data; Big data analytics; Data analysis; Deep learning; Disease diagnosis; Feature selection; Health care; Internet of Things; Machine learning; Map reduce; Mathematical analysis; Particle swarm optimization; Swarm intelligence; 5G networking; Big data analytics; Feature selection; Disease diagnosis; Map reduce
• ISSN: 0045-7906; 1879-0755
• DOI: 10.1016/j.compeleceng.2021.107376

•Propose an optimal Deep Learning Approaches and Healthcare Big Data Analytics for Mobile Networks in 5G environment•Develop a new big data analytic-based feature selection and DBN based disease diagnosis model.•Design a Link-based Quasi Oppositional Binary Particle Swarm Optimization (LQOBPSO) algorithm for feature selection. Recent developments in wireless networking, big data technologies like 5G networks, healthcare big data analytics, Internet of Things (IoT) and other advanced technologies in wearables and Artificial Intelligence (AI) have enabled the development of intelligent disease diagnosis models. The current study devises a new big data analytic-based feature selection and Deep Belief Network (DBN)-based disease diagnostic model. To reduce the number of features and curse of dimensionality, a Link-based Quasi Oppositional Binary Particle Swarm Optimization Algorithm is used in feature selection to narrow down an optimal set of features. The application of quasi-oppositional mechanism in BPSO algorithm helps in increasing the convergence rate. Followed by, the DBN model is applied as a classifier to diagnose the existence of disease from feature-reduced data. A series of experiments was conducted to emphasize the performance of the presented model. The obtained experimental outcomes showcased that the presented model yielded better results in several aspects. [Display omitted]