Effective ensemble learning approach for large-scale medical data analytics

Abstract

In the medical field, several resources of big data like patient medical reports, hospital information, information on diagnostic tests, and data on IoT (Internet of Things)-based smart medical devices form a vast asset to future analysis. Big data produced and recorded in the field of biomedical research are primarily relevant to comprehensive health services. Due to the enormous quantity of storage, appropriate administration, and computing resources, big data demands the most effective and advance data analytical tool. Thus, most of the robust algorithms or tools used to get insights from large datasets, especially to analyze, are from the Machine Learning (ML) domain. Numerous academicians and researchers have suggested ML techniques to medical-oriented data analytics in recent times to increase the accuracy of assessments. Though most of the existing approaches were employed independently and retrieved an acceptable outcome, they still lacked optimal accuracy. To improve accuracy, this article presents an ensemble learning strategy wherein the outcomes of two ML algorithms, namely 'Edge Detection Instance preference (EDIP)' and 'Extreme Gradient Boosting (XGboost)' are integrated. For aggregating precision, the voting technique is utilized. As a consequence of these findings, it is determined that the proposed learning technique achieves the highest level of accuracy when compared with the existing methods. The ensemble learning attains a better accuracy rate for all the four tested datasets, namely, blood cancer (98.43%), liver cancer (97.47%), heart disease (96.34%), and diabetes (98.14%).