Using Neural Networks in the Identification of Signatures for Prediction of Alzheimer's Disease

Abstract

Alzheimer's disease (AD) is now considered the most common type of dementia in the population. Although it is a degenerative and irreversible disease, if diagnosed early, medications may be administered to slow the progression of symptoms and provide a better quality of life for the patient. Herbert et al. And Gòmez conducted studies with classifiers contained in the software Weka using a database with values of 120 blood proteins, and they noticed that they could classify the patient may or may not be diagnosed with AD with an accuracy rate of 93% and 65%, respectively. Thus, this study aims to use neural networks such as Multi-layer Perceptron, Extreme-learning Machine and Reservoir Computing to perform early diagnosis of a patient with or without AD and Mild Cognitive Impairment (MCI), another common type of disease. This article also envisions to utilize the Random Forest Algorithm and the feature selection method available on Weka called Info Gain Attribute Eval to select proteins from the original set and, thus, create a new protein signature. Through experiments it can be concluded that the best performance was obtained with the MLP and the new signatures created with the Random Forest achieved better results than those available in the literature.