scBIT: Integrating Single-Cell Transcriptomic Data Into fMRI-Based Prediction for Alzheimer's Disease Diagnosis.

Area of Science:

• Neuroscience
• Genomics
• Computational Biology

Background:

• Functional MRI (fMRI) and single-nucleus RNA (snRNA) are key Alzheimer's disease (AD) research tools.
• Integrating these modalities offers novel insights but is underexplored.

Purpose of the Study:

• Introduce scBIT, a novel method for enhancing AD prediction by integrating fMRI and snRNA data.
• Improve diagnostic accuracy and interpretability in AD research through cross-modal learning.

Main Methods:

• scBIT segments snRNA data into cell-type-specific gene networks.
• A self-explainable graph neural network extracts critical subgraphs for analysis.
• Demographic and genetic similarities pair snRNA and fMRI data for cross-modal learning.

Main Results:

• scBIT significantly improves fMRI-based AD prediction models.
• Enhances binary classification accuracy by 3.39% and five-class classification by 26.59%.
• Reveals intricate brain region-gene associations and aids biomarker discovery.

Conclusions:

• scBIT advances brain imaging transcriptomics to the single-cell level for AD research.
• Offers a powerful, interpretable method for integrating multi-modal data in neurodegenerative disease studies.
• Facilitates robust biomarker discovery by linking neural function with molecular mechanisms.