Questions about EEG channel labels and demographics label distribution
Junhao Zhang
Dear PhysioNet Challenge 2026 Team,
I am working with the official challenge data for our student project, and we have two clarification questions.
First, regarding channel_table.csv, we noticed that some EEG aliases are listed as explicit derivations, such as F3-M2, C3-M2, F4-M1, or C4-M1, while some aliases appear as single-electrode labels, such as F3, F4, C3, C4, O1, or O2.
Could you clarify how these single-electrode EEG labels should be interpreted?
Second, we noticed a substantial change in the Cognitive_Impairment label distribution between an earlier demographics file and the latest demographics file. The earlier file contained 622 records with 311 True and 311 False labels, while the latest file contains 1103 records with 84 True and 1019 False labels.
Could you clarify whether this imbalance in the latest demographics file is intentional? Should teams treat the latest demographics file as the source of truth for training and validation design? We would also like to know
whether the earlier balanced file should be understood as an earlier release, development subset, or otherwise superseded version.
Thank you for your guidance.
I'm sending you this message now because it's a good time for me. I do not expect you to read, respond, or action if it is outside your work hours.