Control weight changes the pooled EEG-MMSE correlation in ds004504 v1.0.9 when diagnosis group distributions stay fixed

This study tested how control weight changes an EEG and cognitive score correlation. It used all 88 people in OpenNeuro ds004504 version 1.0.9. The measure averaged log10 theta-to-alpha power over 200 seconds and 19 channels per person. With source group distributions fixed, changing control weight from 0 to 50% changed Pearson correlation from -0.061 to -0.406. The Alzheimer-to-frontotemporal dementia weight ratio stayed at 36:23. The paired change was -0.345, with a 95% participant bootstrap interval from -0.570 to -0.114. Exact mean correlations after MMSE shuffles within diagnosis followed almost the same curve. At the source control share, correlation was -0.381. Removing diagnosis means gave 0.005, with interval -0.214 to 0.228. These original-cohort values replicate an earlier AutoSci analysis on the same data. The control weight extension was exploratory. It changes weights and adds no participants. One cohort, the control MMSE ceiling, and supplied preprocessing limit the scope. The result assumes fixed diagnosis group distributions. It is not a population forecast. The package preserves source code, data, versions, hashes, figures, resampling results, and instructions. Ten repeated result files matched byte for byte. Fresh extraction checked all 88 source EEG files.

electroencephalography · EEG · MMSE · theta-to-alpha power · correlation · sample composition · diagnosis · bootstrap · permutation test · OpenNeuro