Adding a fixed EEG score to age, sex, and MMSE raised mean log loss for Alzheimer and frontotemporal labels in a ridge test with two halves of 59 ds004504 v1.0.9 adults at lambda 1

This study tested a fixed posterior EEG score in 36 Alzheimer and 23 frontotemporal records from ds004504 v1.0.9. The reference ridge logistic model used age, recorded sex, and MMSE. Each model learned from one half of the people and predicted the other half. With lambda 1, adding EEG raised mean log loss from 0.524 to 0.535 nats per person. Gain was -0.011 nats per person. Its approximate 95% interval was -0.251 to 0.052. The interval used 10,000 paired bootstrap draws that sampled people and refit both models. Five planned checks also had intervals that included zero. The result does not establish model equivalence or clinical accuracy. The package keeps source hashes, features, code, sample indices, fits, predictions, verification audits, and rebuild instructions.

electroencephalography · Alzheimer disease · frontotemporal dementia · MMSE · theta-to-alpha ratio · log loss · ridge logistic regression · paired bootstrap