Linear age and gender models show higher EEG theta-to-alpha ratios in Alzheimer disease than controls in both 100-second ds004504 v1.0.9 samples

We tested EEG theta-to-alpha power ratios in 88 people from OpenNeuro ds004504, version 1.0.9. Each person supplied one recording and two samples of 100 seconds of retained EEG. We averaged log10 ratios across 19 channels and ten epochs per sample. Under a linear age and gender model, the Alzheimer–control multiplier was 1.49 in the early sample and 1.50 in the late sample. The estimate from both samples was 1.49, with a 95% HC3 interval of 1.22 to 1.83. Concordance on log10 ratios was 0.918 in Alzheimer disease, 0.711 in controls, and 0.807 in frontotemporal dementia. Pooled concordance rose from 0.473 for 10-second samples to 0.842 for 100-second samples. Two controls had at least a twofold change between the longer samples. The package contains the analysis data, paired bootstrap indices, source hashes, code, figures, and repeat instructions. Eleven fresh source extractions matched the archived features. These results describe variation within one recording. They do not establish reliability across sessions or clinical diagnosis accuracy.

electroencephalography · EEG · theta-to-alpha ratio · concordance correlation coefficient · Alzheimer disease · frontotemporal dementia · measurement agreement · OpenNeuro