A Hybrid Machine Learning Framework For Biomarkers Based ADNI Disease Prediction

Abstract

In most of the real-time applications, machine learning algorithms are used to predict the Alzheimer’s disease on high dimensional feature space. However, the condition of Alzheimer Dementia (AD) exponentially progresses due to lack of early intervention. Most
of the traditional ADNI models are independent of image feature space and biomarkers due to high computational time and memory. In order to improve the disease prediction rate, this research work use multiple biomarkers for disease prediction on the ADNI training
data. In this work, an improved CNN based feature selection method, a segmentation model and classification model are implemented on the large number of feature space and biomarkers. Current algorithms are tested and evaluated; an improved set feature selection
method is proposed with re-sampling strategies. Experimental results proved that the present CNN feature selection-based segmentation and classification model has better prediction rate than the conventional models on high dimensional features.