Distinguishing Alzheimer's disease from other dementias using pattern profile analysis in the Meyers...

This exploratory study aimed to assess the efficacy of pattern-matching statistical methods within the Meyers Neuropsychological Battery (MNB). It compared neuropsychological test data profiles of Alzheimer's disease (AD) patients from three independent samples against four MNB dementia groups: MNB-AD, MNB-Vascular Dementia (VaD), MNB-Dementia with Lewy bodies (DLB), and MNB-Parkinson's disease dementia (PDD).

Three AD-independent samples completed either the MNB (referred to as I-MNB-AD), Dementia Rating Scale-2 with additional testing (denoted as DRS-Plus-AD), or the Repeatable Battery for the Assessment of Neuropsychological Status (designated as RBANS-AD). Test data profiles were cross-validated with four MNB dementia comparison group datasets. Statistical methods included Pearson correlation, Kullback-Leibler (KL) divergence, pooled effect size (Cohen's d), Configuration, and MNB Code.

Classification accuracy ranged from 40% (Pearson r) to 88% (Cohen's d) in the I-MNB-AD sample, 47% (Cohen's d) to 93% (KL) in the DRS-Plus-AD sample, and 47% (Pearson r) to 78% (Configuration) in the RBANS-AD sample. Some methods showed limited effectiveness depending on the sample and comparison group analyzed, while others demonstrated strong performance. Using a simple majority count of agreement, classification rates for selecting the MNB-AD comparison group were 80% (I-MNB-AD), 85% (DRS-Plus-AD), and 66% (RBANS-AD).

This exploratory study demonstrates that specific statistical methods employed in the MNB for pattern-matching analysis effectively differentiated neuropsychological profiles of individuals with AD from other types of dementia, contributing to improved diagnostic precision. The findings underscore the potential advantages of pattern-matching analysis, advocating for further research to validate and refine its application.