Stratifying Mild Cognitive Impairment Patients via Embedded Transcriptomic Data and Clustering Analysis.

Area of Science:

• Neuroscience
• Genomics
• Computational Biology

Background:

• Alzheimer's disease (AD) is a progressive neurodegenerative disorder with significant mortality.
• Mild cognitive impairment (MCI) is a risk factor for AD dementia, but current diagnostic methods lack molecular precision.
• Precision medicine aims to classify pathologies using molecular markers for improved early detection.

Purpose of the Study:

• To develop a molecular stratification method for MCI patients using blood transcriptomic data.
• To identify distinct MCI patient clusters based on molecular profiles.
• To improve early and precise diagnosis of progressive dementia and AD.

Main Methods:

• Blood transcriptomic data analysis.
• Unsupervised and supervised feature selection from gene expression data.
• Deep learning-based autoencoders for data embedding into a latent space.
• Clustering algorithms applied to the latent representation for patient stratification.

Main Results:

• Identified distinct MCI patient clusters with significant molecular, clinical, and pathophysiological differences.
• Demonstrated differences between MCI clusters, AD patients, and control samples.
• Established a molecular basis for classifying MCI, aiding in early AD detection.

Conclusions:

• Clustering-based molecular stratification of MCI patients offers an objective diagnostic approach.
• This method can lead to earlier and more precise diagnosis of progressive dementia.
• Blood transcriptomic analysis provides a powerful tool for understanding and classifying neurodegenerative conditions like AD.