ABSTRACT

In the current era of technological evolution, almost every aspect of human life has been digitalised through the use of modern technologies. Consequently, medical science, human healthcare technologies, medical practices and research are concerned with evolutionary technologies for better accuracy and efficiency in detecting diseases as well as other similar tasks. This exponentially growing data plays a negative role while manipulating data sets for medical purposes and research. These data demonstrate structured, unstructured, noisy and irrelevant types of data, which seem impossible to analyse manually. The emerging medical technologies including sensors, actuators, smart devices, cloud computing, and smart watches are generating a large number of data which are almost impossible to process for gaining knowledge using traditional approaches. In healthcare systems, the immediate exploration of this data is mandatory to determine patients’ current health status and for accurate diagnoses. Thus, evolutionary approaches along with flourished pre-processing algorithms are mandatory to revealed specific information and for further decision-making. Realizing the potential worth of these data we embed an improved pre-processing mechanism with an evolutionary MapReduce approach to analyse data after the successful management of noisy, irrelevant data. We employ this approach on medical patients’ data. We compare between a pre-processing based MapReduce approach and pre-processing less MapReduce approach. Also, we also compare a pre-processing approach based on mapping only with a pre-processing based MapReduce as well as Pre-processing less MapReduce. These comparisons are demonstrated theoretically, mathematically, and graphically. In addition, comparisons are made for various sizes of data sets. For clarification, we also measure the accuracy, specificity and sensitivity of the algorithms we have considered here.