Five minutes of calibration selected depth contacts with higher later sound correlations in nine Podcast recordings with a common average reference
A short early interval can set a depth-contact list before a later sound model is tested. We tested this method in all nine participants of OpenNeuro ds005574, version 1.0.2. The analysis used 325 depth contacts that passed the source metadata rules. The first five minutes selected the highest-scoring 20 percent of contacts within each participant. Models fitted to that interval were tested in minutes 20-30 of the same story. The selected contacts had a mean test correlation of 0.182. The mean across all eligible contacts was 0.053. The mean participant gain was 0.129, with a 95 percent bootstrap interval of 0.081-0.175. All nine gains were positive. Eight participants passed a sound time-shift control. Mean selected-contact R-squared was -0.007. The evidence supports correlation-based contact screening within this recording and reference system. It does not establish stable amplitude prediction or transfer to a new story.
depth contacts · high-gamma activity · sound encoding · ridge regression · contact selection · intracranial electrophysiology
References
2- usesdoi:10.18112/openneuro.ds005574.v1.0.2
The analysis used the released high-gamma data and sound features from OpenNeuro ds005574 version 1.0.2.
- doi:10.1038/s41597-025-05462-2
Cited by
6 publicationsPublic catalog entries on Sciscape.
- usesSound features gave an uncertain gain beyond signal history after five minutes of training in nine Podcast SEEG recordings with a common average reference
This publication supplied the ten compact public input arrays, original source manifest, integral binning recipe, contact audit, and source metadata. We checked all input archive hashes. We added a recent same-contact history baseline, used all 325 eligible contacts, and tested added sound prediction value. This is a dependent analysis of the same cohort and story.
- usesSound band differences gave no clear mean R² gain in nine depth recordings with a common average reference and fixed ridge training
We reused the ten compact input arrays and checked source provenance. We adapted the source retrieval code. The new comparison uses level-only contact selection and tests incremental sound-band prediction.
- usesBipolar subtraction reduced mean later sound prediction correlation for depth pairs selected with a common average reference in nine Podcast recordings
The parent package supplied the compact sound input, early model and time rules, and copied source metadata. Its released CAR amplitude bins enabled numerical source checks. The sound input was independently rebuilt. Both new neural targets were calculated from the pinned cleaned voltage files.
- usesSound reduced mean test error at selected depth contacts in nine Podcast recordings with common average reference and controls for contact history and other shafts
Used its original compact data lineage and exact 100 ms integration method, preserved in the later package.
- usesEarlier sound had a mean test R2 advantage of 0.0082 over later sound at 68 selected SEEG contacts in nine Podcast recordings
Its earlier public source extraction supplied the compact-data lineage and exact 100 ms integration method used here.
- usesSound reduced mean later error by 4.8 percent at selected depth contacts in nine Podcast recordings with common average reference and other-shaft mean control
We used its ten checked compact data files, source manifest, exact bin integration, and early contact selection method. We reproduced its contact list and sound-only correlations. We added neural controls that exclude the target shaft and tested added sound prediction value on the same later segment.