Association rule learning in neuropsychological data analysis for Alzheimer's disease

Area of Science:

• Neuroscience
• Computational Psychology
• Data Mining

Background:

• Clinicians require more efficient raw data analysis methods in neuropsychological assessments.
• Current analytical tools for raw neuropsychological data are limited, hindering detailed clinical insights.

Purpose of the Study:

• To evaluate association rule learning, specifically Frequent Pattern Growth (FP-Growth), as a tool for analyzing neuropsychological raw data.
• To assess the clinical utility of fine-grained analysis and the feasibility of predicting response patterns in Alzheimer's disease (AD) using FP-Growth.

Main Methods:

• Utilized FP-Growth algorithm to mine patterns from the Consortium to Establish a Registry for Alzheimer's Disease Neuropsychological Battery (CERAD-NB) database.
• Analyzed data from 84 confirmed AD cases and 294 controls at baseline and one-year follow-up.
• Focused on identifying frequent itemsets and predictive association rules within raw score data.

Main Results:

• FP-Growth successfully identified discernable patterns in frequent itemsets across different AD severity groups (controls, mild, moderate to severe) (p < .001, η² = .488).
• The analysis demonstrated that patterns within raw data scores are differentiable across clinical and control groups.
• Frequent itemsets and predictive association rules were successfully generated, indicating group differentiation.

Conclusions:

• FP-Growth is a viable supplementary analysis tool for neuropsychological assessment.
• The method provides an additional layer of data analysis and predictive capabilities for item responses.
• FP-Growth aids in clinical decision-making by offering finer analysis of raw neuropsychological data.