Observed arrival · 2026-10-11
An EEG depression baseline meets a batch-effect trap
A project site documents a reproducible EEG depression-classification baseline and the methodological problems found while testing it.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- EEG researchers and machine-learning practitioners
- Worth noticing
- The project says a control using no EEG signals showed acquisition batch could explain most of the reported subject-level performance.
Field notes
The homepage describes a baseline built with MNE preprocessing, band-power features, and an SVM, reporting subject-level results across 62 participants and 35,400 segments. Its posts explain why participant-level splitting matters: adjacent two-second segments from one person can otherwise land in both training and test sets, allowing identity leakage. The project also lists deep-learning and Mamba work as a next step, not a completed capability.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
◎NicheUnusually specific use
One card from the complete issue