Deep Learning-Based Prediction of Alzheimer's Disease Using Microarray Gene Expression Data.

Area of Science:

• Genomics and Bioinformatics
• Computational Neuroscience
• Artificial Intelligence in Medicine

Background:

• Alzheimer's disease (AD) is a genetically complex neurodegenerative disorder.
• Microarray technology offers insights but faces challenges due to high dimensionality and small sample sizes.
• Gene selection is crucial for improving diagnostic models for AD.

Purpose of the Study:

• To investigate deep learning techniques, specifically Convolutional Neural Networks (CNNs), for predicting Alzheimer's disease using gene expression data.
• To develop a reliable predictive model for early AD detection and diagnosis.
• To enhance precision medicine approaches for neurodegenerative disorders.

Main Methods:

• Applied gene selection techniques, including Principal Component Analysis (PCA) and Singular Value Decomposition (SVD), to reduce dimensionality in microarray datasets.
• Utilized a seven-layer Convolutional Neural Network (CNN) architecture for classification.
• Trained and evaluated PCA-CNN and SVD-CNN models on an Alzheimer's disease dataset.

Main Results:

• The PCA-CNN model achieved 96.60% accuracy with a loss of 0.3503.
• The SVD-CNN model demonstrated superior performance, reaching 97.08% accuracy with a loss of 0.2466.
• Both models highlighted the effectiveness of gene selection in enhancing classification accuracy.

Conclusions:

• Integrating gene selection methods with deep learning architectures offers a powerful framework for improving Alzheimer's disease prediction.
• The developed models show significant potential for early AD diagnosis and intervention.
• Future research will explore alternative gene selection techniques and deep learning architectures for broader applications.