Identifying Key MicroRNA Signatures for Neurodegenerative Diseases With Machine Learning Methods

Area of Science:

• Biomedical Informatics
• Neuroscience
• Genomics

Background:

• Neurodegenerative diseases like Alzheimer's (AD) and vascular dementia (VaD) cause cognitive decline, but their exact causes are unknown, and diagnosis is challenging.
• MicroRNAs (miRNAs) are implicated in various biological processes, and their dysregulation is linked to neurodegenerative conditions.

Purpose of the Study:

• To develop a machine learning model for classifying normal controls (NC), AD, VaD, dementia with Lewy bodies, and mild cognitive impairment.
• To identify specific miRNA expression profiles associated with different types of dementia and cognitive impairment.

Main Methods:

• Utilized Boruta, minimum redundancy maximum relevance, and Monte Carlo feature selection for miRNA feature selection.
• Employed incremental feature selection with random forest and PART classifiers to identify optimal feature subsets.
• Extracted classification rules from the best-performing PART models.

Main Results:

• Successfully distinguished between NC, AD, VaD, and mild cognitive impairment at the miRNA expression level.
• Identified key miRNA candidates, including hsa-miR-3184-5p, has-miR-6088, and has-miR-4649, associated with neurodegenerative diseases.
• Validated findings with existing research, confirming the role of specific miRNAs in cognitive impairment.

Conclusions:

• The developed machine learning technique offers a promising avenue for diagnosing neurodegenerative diseases and cognitive impairments.
• Specific miRNAs hold potential as diagnostic biomarkers and targets for understanding the pathogenic mechanisms of dementia.
• Further research into miRNA's role in neurodegeneration can advance diagnostic and therapeutic strategies.