Chronic Kidney Disease Prediction using Hybrid Optimization Search Model (HOSM)

Abstract

Kidney is an important organ of the human body helps to keep balancing the functionality of all other organs. Chronic kidney diseases(CKD)create many impacts in recent days due to changes in life style habits, lack of health awareness and irregular treatments on health issues. Diabetics is one of the largest impacted diseases affect the humans. A chronic diabetic allows slow degradation of organs health when untreated regularly. A chronic disease covers various health problems that is resilient in diagnosis. The proposed research work is aims to develop an comparative analysis model that deep dig the attributes of the chronic kidney dataset collected from Kaggle.com. the standard dataset act as the benchmark for initially creating the machine learning models. The novel framework acted upon with tuning the hyper parameters using hybrid optimization approach. Machine learning models such as Adaptive boosting model (ADAB), CatBoostAlgorithm (CA), K-Nearest neighbor algorithm (KNN), Probability boosted regression (PBR) and Wide – Deep decision tree model (WADET) etc. the comparative analysis on CKD is validated through output parameters in the form of prediction accuracy, system precision, Recall and F1 score. The novel system relies on hybrid optimization enabled search model represented as HOSM to find out the best adaptive approach on developed systems. The robust framework is iteratively performing the validation process thus provides N-Cross validation mechanism that adopt dynamic inputs. The presented system using HOSM-WADET achieved 98.77% cumulative accuracy comparing the state of art approaches.