Analyzing Healthcare Data Using Water Wave Optimization-Based Clustering Technique

Introduction

In today's world, an enormous amount of healthcare data is generated through medical instruments, patients care, and hospitals. This data contains various hidden patterns and important information for better treatment of patients and curing the people of diseases. So, at present time, healthcare data analysis has become an important activity for achieving good medical care decisions. It is seen that incorrect decisions may lead to delays in medical treatment and sometimes, human life can be lost. In literature, several medical decision systems, expert systems, and diagnostic systems have been reported for better treatment, analysis, and diagnosis (Bahrami & Shirvani, 2015; Haraty & Masud, 2015; Ahlqvist et al., 2018). Further, biomedical experts describe the medical diagnosis and analysis as a decision-oriented process based on historical data. It is noticed that computer-aided, and automatic diagnostic systems are more superior to structural computing system. Because these systems can process large amounts of data in a short span and provide more accurate results. Further, these systems having the capability to train the model using similar data or cases. So, due to fast processing and better learning capability, quick decisions can be drawn for better treatment and saving human life. These systems also help physicians and specialists by minimizing the diagnosis error. The other benefits of these systems are improved diagnosis accuracy, reduced diagnosis time, and treatment cost (Santhanam & Padmavathi, 2015; Kefelegn & Kamat, 2018; Guo et al., 2020). But it is observed that several deficiencies are associated with these systems such as accuracy, attributes selection for disease diagnosis, biased decisions, misinterpretation of diagnostic results, etc. On the other side, it is seen that computational intelligence (CI) is widely used in the medical field for either analyzing or diagnosis of healthcare data and disease prediction. CI contains different ideas, models, procedures that are inspired through various natural phenomena, law, and swarm behaviors. CI methods are widely adopted for solving complex real-world difficulties. Some CI algorithms are listed as Water Cycle Algorithm, Water Flow Algorithm, Artificial Chemical Reaction Optimization Algorithm, Artificial Chemical Process, Artificial Bee Colony, Bat Algorithm, Beehive, and Base Optimization Algorithm. It is noted that CI methods can be used for preprocessing of medical data such as data cleaning, missing value imputation, attribute selection, attribute weighting), classification, clustering, and prediction of disease. The numerous diagnostic models are reported in the literature for disease diagnosis based on CI methods (Altayeva & Cho, 2016; Ni et al., 2017; Nilashi et al., 2018; Jothi & Husain, 2015). But diagnostic accuracy is still a challenging area, especially for healthcare datasets. Recently, water wave optimization gains wide popularity among the research community and obtains optimal results for numerous optimization problems; (i) constrained and unconstrained optimization (Lenin et al., 2016; Siva et al., 2016) (ii) scheduling (Shao et al., 2018; Zhao et al., 2019), (iii) allocation of the frequency spectrum (Singh et al., 2019), (iv) multi-objective optimization (Hematabadi & Foroud, 2019), etc. The performance of the WWO algorithm is affected due to a lack of balance among search mechanisms and in turn, converges on premature solutions.