Cognitive biomarker prioritization in Alzheimer's Disease using brain morphometric data.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Biomedical Informatics

Background:

• Cognitive assessments are crucial for Alzheimer's Disease (AD) diagnosis.
• Current cognitive test selection lacks individual customization, despite numerous tools and time constraints.
• Personalized prioritization of cognitive assessments is needed for efficient AD diagnosis.

Purpose of the Study:

• To develop a machine learning paradigm for personalized cognitive assessment prioritization.
• To enable customized selection of cognitive tests for individual subjects.
• To enhance the efficiency of cognitive assessments in clinical settings.

Main Methods:

• Adaptation of a learning-to-rank approach to prioritize cognitive assessments.
• Development of a scoring function to rank the effectiveness of cognitive tests.
• Extension of the learning-to-rank method for improved separation of effective and less effective assessments.

Main Results:

• The proposed machine learning paradigm significantly outperforms state-of-the-art baselines.
• Empirical studies on ADNI data show superior performance in identifying and prioritizing individual-specific cognitive biomarkers.
• Improvements of up to 22.1% and 19.7% were observed in prioritizing cognitive features.

Conclusions:

• The developed paradigm demonstrates superior performance in prioritizing cognitive biomarkers.
• Prioritized cognitive biomarkers can aid in personalized AD diagnosis and disease subtyping.
• This approach holds potential for advancing precision medicine in Alzheimer's Disease.