BIG DATA ANALYTICS FOR HEALTHCARE

Abstract

The rapid development of urbanization improved our daily life but also leads to a series of urban diseases. Worldwide around 2.4 million deaths could be prevented if a practice of proper hygiene can be maintained. Nowadays the focus of the government is to provide better and safe health to its citizens. So many new policies are coming into existence and a large number of funds have been assigned to implement these policies. The Indian government has recently announced the world’s largest health care program under the Ayushman Bharat scheme. In this Program Policymakers and business practitioner around the world extending their extraordinary efforts in the field of e-health. Many local people awareness programs are also running which provides information to the citizens about infectious diseases like AIDS, Swine flu, Malaria, Covid-19, etc. These awareness programs run under the supervision of local health authorities. Since the digitization makes the work easier to spread awareness among the people. The primary requirement of a government is to provide a healthy environment and it can be achieved through digitalization. The development of computer science and regional information generates a vast amount of data and creates an environment for disease prediction. Big data analytics (BDA) is a revolution in information technology and can be used for healthcare sector to perform data analysis. As the data is accumulated from various sources such as wearable sensor technology, smart phone, robots, IoT, etc. This data can be used for clinical predictions by using emerging machine learning algorithms and BDA techniques. BDA provides systematic information based on the vast amount of healthcare data to develop better healthcare system. With the help of the analyzed patterns, valuable information can be extracted and used by the policymakers to build a protective environment for the better healthcare system. Therefore, the present research work integrates data management scheme for healthcare along with data analysis system to achieve the target towards health 4.0 for providing better healthcare service to the individual users. This research aim is to incorporate the five significant contributions to the healthcare industry. v First, the literature review of big data analysis has been performed for healthcare technology to highlight the challenges of the healthcare industry and identify the various mechanism to overcome these challenges. Second, we proposed a novel healthcare multi-phase architecture (HCMP) to predict chronic kidney disease. The HCMP architecture works on six different layers namely: data-collection, data-storage, data-management, data-processing, data-analysis, and report-generation. The data-storage and data-management layers were performed on heterogeneous Hadoop cluster and the profiling methods were used to consider three situations for calculating the capacity ratio of each DataNode in the cluster. MapReduce is used for parallel data processing. Furthermore, horizontal scaling is performed in the Hadoop cluster, and the performance of every DataNode is investigated based on a capacity ratio. The data-analysis layer has performed classification tasks using a Decision tree, K Nearest Neighbors (KNN) classification, Kernel distributed Naïve Bayes, Simple distributed Naïve Bayes, Random Forest, and Random tree. Among these classifiers, Kernel distributed Naïve Bayes has produced the best results. The experiments are performed using tools such as Hadoop and RapidMiner to evaluate, analyze the efficiency and performance of the proposed architecture. Third, the HCMP architecture also deployed the proposed MySymptom algorithm to filter the Chronic Kidney Disease (CKD) dataset of patients according to their symptoms at the data processing layer. The case study of CKD in Indian environment has been explored. Forth, the proposed work is enhanced to analyze multiple disease using a new Hadoop-based Optimal Healthcare Classification Multi-Disease Diagnostic (OHCMDD) architecture for handling multiple diseases database with reduced features set. The proposed OHCMDD architecture also handles the DataNode deletion problem in Hadoop’s cluster. Also, an intelligent classification prediction model is introduced for healthcare, namely Density-Based Features Selection with Spider Monkey Optimization (D-SMO) is used for chronic kidney disease (CKD), heart and diabetes disease. The empirical results of the D-SMO algorithm are compared with existing methods. The presented intelligent multi-disease model outperformed the other methods and implemented using the Hadoop cluster, HDFS and R platform. Lastly, the proposed e-health architecture is deployed to achieve the target of health 4.0. The future of health management will become timelier and more personalized. As new technologies will empower individuals to conduct their health monitoring by using cyber–physical systems. The design principles of industry 4.0 connect the physical and virtual world in real-time. vi Virtualization in health happens after the emergence of Information and Communication Technologies (ICT). For this 5G, the next-generation mobile network provides ambient intelligence for orchestration of medical services so that government and private companies can reconsider health prospects. These technological developments in healthcare, big data and industry 4.0 are individually helping to achieve the goal of health 4.0. The proposed work also provides the road map for health 4.0 along with different technologies involved in it. The analysis and performance evaluation of the experimental results demonstrate that the proposed work provides a reliable architecture for better healthcare environment. Moreover, the comparative analysis displays that the proposed work showed an improvement. Thus, the proposed study provides an optimal and effective architecture for healthcare industry towards health 4.0.