A Robust Uncertainty Based Animals Migration Optimization Based Clustering for Alzheimer Disease Prediction

Abstract

In the advanced environment the health information is maintained and the diseases are predicted using machine learning models. One of the primary neurodegenerative disorders which affects the life quality of the elderly persons is Alzheimer, diagnosing its presence in its earlier stages may help the medical experts to slow down the aggressiveness of the disease. In this paper, a novel uncertainty-based clustering model is developed to handle the vagueness in selection of centroids to overcome the problem of outliers and noisy instances which affect the performance of the prediction model. The unknown patterns are very challenging while using the unsupervised learning algorithms, hence in this work an uncertainty-based optimization algorithm is used to handle the unknown pattern of Alzheimer Disease (AD) patients. Initially, the instances in AD dataset is converted to the membership value of the dependent variable to exactly define the belongingness of them as AD patterns or non-AD patterns. To overcome the outliers during the process of similar patterns clustering, in this proposed work animal migration optimization algorithm is induced based on their migration behavior, the best instances are selected as centroids while a new instance is considered for clustering. Each instance is validated based on their fitness value and the instances with best fitness value is considered as centroids. The fuzzy euclidean distance is used for handing the uncertainty in handling the outliers. The simulation results on OASIS dataset proved that the proposed uncertainty-based Animal Migration optimization algorithm (UAMO) produced better results compared to other state of arts clustering models.