Blood biomarker-based classification study for neurodegenerative diseases.

Area of Science:

• Biomedical Informatics
• Neuroscience
• Computational Biology

Background:

• Neurodegenerative diseases like Alzheimer's (AD) and Parkinson's (PD) are increasing with aging populations.
• Early diagnosis and screening are critical for effective treatment and clinical trials.
• Gene expression profiling offers a powerful tool for identifying disease biomarkers.

Purpose of the Study:

• To identify blood-based biomarkers for Alzheimer's disease (AD) and Parkinson's disease (PD) using machine learning.
• To compare the efficacy of traditional machine learning with deep learning approaches for biomarker discovery.

Main Methods:

• Applied five machine learning (ML) approaches with feature selection to gene expression data.
• Developed optimal Random Forest (RF) models for AD and PD biomarker identification.
• Evaluated deep learning models, specifically Convolutional Neural Networks (CNNs), for biomarker detection.

Main Results:

• An optimal RF model for AD identified 159 gene markers with a ROC AUC of 0.886.
• An optimal RF model for PD achieved a ROC AUC of 0.743.
• CNNs demonstrated consistent performance on both AD (ROC AUC = 0.810) and PD (ROC AUC = 0.715) datasets.

Conclusions:

• Machine learning, particularly Random Forest, is effective in discovering blood-based gene expression biomarkers for AD and PD.
• Deep learning methods like CNNs show promise for future biomarker detection in neurodegenerative diseases.
• These findings support the use of gene expression data and ML for early detection and improved management of AD and PD.