Dementia-related user-based collaborative filtering for imputing missing data and generating a reliability scale on...

Area of Science:

• Neurology
• Medical Informatics
• Artificial Intelligence

Background:

• Dementia diagnosis can be challenging due to missing or incorrect clinical test scores.
• Morphometrics and demographics are vital for accurate dementia assessment.
• Existing clinical data may lack reliability, impacting diagnostic and prognostic accuracy.

Purpose of the Study:

• To develop a decision support system for improved dementia diagnosis and prognosis.
• To impute and verify clinical test scores using brain MRI analysis and demographics.
• To enhance the understanding and reliability of dementia diagnostic tools.

Main Methods:

• Unsupervised dementia-related user-based collaborative filtering was employed to impute missing clinical scores.
• Brain MRI analysis and demographic data were integrated with clinical scores.
• A reliability scale was developed based on succession rates to assess score consistency.
• Analysis of 816 ADNI1-screening samples with 603 features was performed.

Main Results:

• The proposed imputation system effectively minimized errors in clinical test scores.
• Certain collaborative filtering configurations demonstrated superior performance compared to other imputation methods.
• The developed reliability scale showed promise in evaluating the consistency of existing clinical data.
• The system successfully integrated MRI, demographics, and clinical data for enhanced analysis.

Conclusions:

• The developed imputation system and reliability scale offer a promising approach to support clinical dementia testing.
• This methodology can aid medical doctors in making more accurate diagnoses and prognoses.
• The decision support system provides an easy-to-understand framework for complex dementia assessment.