A. Kaliappan and D. Chitra. Kringing Regressive Map reduce Entropy Feature Extraction based Rocchio Adaptive Boost Ensemble Classifier for Early Disease Diagnosis with Big Data. Dynamic Systems and Applications 30 (2021) No.6, 964-980

https://doi.org/10.46719/dsa20213064

ABSTRACT. 
With big data growth in biomedical and healthcare communities, preciseinvestigation of medical data benefits early disease detection, patient care and community services. Conversely, analysis accuracy is reduced when medical data quality is in adequate. Furthermore, different regions exhibit unique characteristics of certain regional diseases, which may deteriorate disease outbreaksprediction .In healthcare industry, several sources for big data comprises of hospital records, patientsmedical records, medical assessmentsoutcomes, and so on. But, big data analytics of health care is still a challenging and time-demanding task. For big data analytics accuracy improvement, a novel technique called Kringing Regressive Map-reduce Entropy Feature Extraction based Rocchio Adaptive boost Ensemble Classification (KRMEFE-RABEC) technique is introduced.The main aim of the KRMEFE-RABEC technique is enhancing disease prediction accuracy with minimum time consumption. The proposed KRMEFE-RABEC technique includes three key processes specifically preprocessing, feature extraction, and classification.  Initially, Kringing Regressive Mapreduce preprocessing is accomplished for cleaning and transforming the raw data into a valuable and understandable format to minimize the complexity of the disease diagnosis. Secondly, the Rényi entropy–basedfeature extraction process is carried out to the pre-processed features forfinding the robust features significant to seizures for accurate disease diagnosis. Finally, the Rocchio Adaptive Base class boost ensemble technique is applied to early epileptic seizure recognition with the robust extracted features by constructing the weak learners. Experimental outcomes demonstrate that suggestedKRMEFE-RABEC technique significantly classifies the epileptic seizure classes by means of accuracy, sensitivity, specificity, and accuracy, time complexity with respect to the number of samples. The outcomesobtained show that the proposed KRMEFE-RABEC technique achieves greater accuracy with minimum time complexity than traditional approaches.

KEYWORDS: Big Data, Disease Diagnosis, Kringing Regressive Mapreduce Preprocessing, Rényi Entropy–Based Feature Extraction, Rocchio Adaptive Base Class Boost Ensemble.