Sparse Multi-Response Tensor Regression for Alzheimer's Disease Study With Multivariate Clinical Assessments

Area of Science:

• Neuroimaging
• Neurodegenerative Disorders
• Biostatistics

Background:

• Alzheimer's disease (AD) and amnestic mild cognitive impairment (MCI) are increasing neurodegenerative conditions.
• Neuroimaging is crucial for understanding AD and MCI, with most studies focusing on binary classification.
• Current research is exploring continuous clinical scores for richer diagnostic information, but few studies model multiple scores concurrently.

Purpose of the Study:

• To propose a novel sparse multi-response tensor regression method.
• To jointly model multiple clinical scores and multiple image voxels for enhanced Alzheimer's disease (AD) and mild cognitive impairment (MCI) analysis.
• To improve disease diagnosis and identify relevant brain subregions associated with AD/MCI outcomes.

Main Methods:

• Developed a sparse multi-response tensor regression model.
• Applied the method to simultaneously analyze multiple clinical outcomes and image voxels.
• Utilized the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset for experimental validation.

Main Results:

• The proposed method effectively models multiple clinical scores and image voxels jointly.
• The approach demonstrated enhanced performance in inferring clinical scores and disease diagnosis.
• Identified specific brain subregions highly relevant to Alzheimer's disease (AD) and mild cognitive impairment (MCI) outcomes.

Conclusions:

• The sparse multi-response tensor regression method offers a powerful tool for analyzing complex neuroimaging data in Alzheimer's disease (AD) research.
• Simultaneous modeling of multiple clinical scores and brain regions improves diagnostic accuracy and provides deeper insights into disease pathology.
• This method outperforms existing solutions and aids in identifying key biomarkers for AD and MCI.