Selective diagnostics of Amyotrophic Lateral Sclerosis, Alzheimer's and Parkinson's Diseases with machine learning and miRNA.

Abstract

The diagnosis of neurological diseases can be expensive, invasive, and inaccurate, as it is often difficult to distinguish between different types of diseases with similar motor symptoms. However, the dysregulation of miRNAs can be used to create a robust machine-learning model for a reliable diagnosis of neurological diseases. We used miRNA sequence descriptors and gene target data to create machine-learning models that can be used as diagnostic tools. The top-performing machine-learning models, trained on filtered miRNA datasets for Amyotrophic Lateral Sclerosis, Alzheimer's and Parkinson's Diseases of this research yielded 94, 97, and 96, percent accuracies, respectively. Analysis of dysregulated miRNA in neurological diseases elucidated novel biomarkers that could be used to diagnose and distinguish between the diseases. Machine-learning models developed using sequence and gene target descriptors of miRNA biomarkers can achieve favorable accuracies for disease classification and attain a robust discerning capability of neurological diseases.