Predicting Neurodegenerative Diseases Using a Novel Blood Biomarkers-based Model by Machine Learning

Abstract

This paper presents machine learning based framework to the analysis and modeling of several neurodegenerative diseases by using features from blood-based biomarkers. The proposed approaches can be employed for early detection of Alzheimer`s disease (AD) or Parkinson`s disease (PD). In particular, we applied LDA (linear discriminant analysis) for visualizing the dataset as 2D or 3D scatter plots. Moreover, we constructed various classifiers for several different tasks of classification, and explore the accuracy of these classifiers. Based on our experiments, random forests are in general a very good choice of these tasks considering both the computing time (during modeling and prediction) and the accuracy.