Big data analytics using deep learning techniques with feature selection and classifiers for diabetic disease diagnosis

Abstract

Big healthcare data presents proficient diabetes disease prediction from newlinemedical patient data by extracting the relevant features from the healthcare newlinesystem. In the medical field, big data is utilized for early disease prediction from newlinepatients. During the diabetes disease diagnosis, feature extraction and data newlineclassification is the most considerable task for minimizing the dimensionality of newlinedata. The effective performance of feature extraction helps to diagnose disease newlinewith higher accuracy through removal of immaterial features. After eliminating newlinefeatures as of medical dataset, patient data with relevant aspects are considered newlinefor classifying medical patient information. Classification is process to classify newlinepatient medical data for detecting diabetes disease. Generally, diabetes is a newlinehazardous disease in data growth that causes death. Diabetic patients are mainly newlineaffected with high blood sugar. With the identification of disease symptoms, newlinedisease diagnosis is attained with minimum time. newlineIn recent research works, several feature selection and classification newlinetechniques were employed to classify patient data at an earlier stage. Thus, newlineseveral techniques have introduced for resolving data classification issue for newlinedisease predictive analytics. The input medical data comprises of huge number newlineof features for classifying data as normal or abnormal. The medical data of newlineaccurate detection plays the essential function of predicting disease. However, newlinethe diagnosis of future conditions was a complex issue within the healthcare newlinesector for minimal risk level. To address enhanced disease diagnosis, different newlineproposed methods are developed with a big medical database. Therefore, Kernel newlinePerceptron Feature Selection based Sparse Bayesian Probabilistic Relevance newlineVector Machine (KPFS-SBPRVM) technique, CapsNet framework and newlineRegressive Interpolated Contingency Statistical Feature Selection based newlineMapReduce Deep Belief Network Classifier (RICSFS-MDBNC) technique newlineredesigned to attain higher accuracy in diabete