A Hybrid Filter Wrapper Embedded-Based Feature Selection for Selecting Important Attributes and Prediction of Chronic Kidney Disease

Today’s most significant healthcare problem that is prevailing is the chronic kidney disease (CKD). The disease integrates well-defined patho- physiological process that will be experimental for determining irregular kidney
functions and the glomerular filtration rates. To forecast the disease, different data mining techniques are used to discover the connections between various elements,
which can be utilized to determine the progress and status of CKD. Data is obtained from the patient’s healthcare records. The main purpose of this research is to avail
the Hybrid Filter Wrapper Embedded-Based Feature Selection (HFWE-FS), which will be utilized to select CKD datasets from potential feature subsets. HFWE-FS algorithm integrates the process of filtering, wrapping and embedding algorithms.
The filter algorithms are integrated with reference on certain metrics: Gini index, gain ratio, One R and Relief. The wrapper algorithms via enhanced bat algorithms are purposed to select the analytical features from wide-range CKD sets of data. The embedded algorithms are underpinned, and this depends on the support vector machine (SVM)-t statistic, which selects the analytical features out of the wide- range CKD dataset. The results of the feature selection algorithms are integrated and identified as the HFWE-FS algorithm. The SVM algorithm for the CKD predic-
tion is proposed as a final stage. The database used is taken from ‘CKD’ imple- mented on the MATLAB. The results perceived that the SVM classifier along with HFWE algorithm gets high classification rate when contrasted with other categori- zation algorithms: Naïve Bayes (NB), artificial neural networks (ANNs) and sup- port vector machine (SVM) in CKD completion.