One calibrated SEEG score gave an uncertain recall forecast gain over task calibration in 16 later sessions
Can a small neural correction improve memory forecasts across sessions? We tested 16 patient pairs from public OpenNeuro SEEG data, ds004809 version 2.2.0. We fit one neural score to source task errors. We then fit an offset and slope with 60 early words in each later session. The task reference used the same number of later calibration parameters. The test used the next 180 words per patient, or 2880 words in total. The mean log loss gain was 0.000343 nats per word. The 95% patient bootstrap interval was [-0.006900, 0.008614]. Seven of 16 patients had a gain. The interval permits benefit and harm. It does not show equal performance. This is an exploratory test on reused session pairs. The plan was saved before new fits, but prior cohort results were known. The study tested the reading step of a memory interface. It did not apply stimulation. The package preserves inputs, code, source records, word forecasts, fitted models, figures, and repeat checks.
stereoelectroencephalography · memory recall · forecast calibration · logistic regression · patient bootstrap
References
5- usesAdded SEEG power increased recall forecast loss in all eight later sessions with a fixed logistic reader sci:sha256:5d98ad3a4b35585f15bfec73be164b2de5da77c70912a6e3c5282178fbe7d256
The source task coding, one-second power recipe, and first eight patient session pairs come from this earlier benchmark through the direct parent package.
- A fixed SEEG recall reader with scaling from 60 early words had higher mean loss than task inputs in eight new session pairs at C = 0.01 sci:sha256:cc2e373832a1a405c2992d004e962cfdf49dcfddb914d6b2ba0dbdb2bf23ff97
[Legacy relation: reproduces] The task-offset and joint-reader loss and Brier scores reproduce 32 parent score rows within 7e-13. The new AUC uses an explicit 1e-12 tie tolerance, so this relation does not claim exact AUC reproduction.
- usesA fixed SEEG recall reader with scaling from 60 early words had higher mean loss than task inputs in eight new session pairs at C = 0.01 sci:sha256:cc2e373832a1a405c2992d004e962cfdf49dcfddb914d6b2ba0dbdb2bf23ff97
This study uses all 64 feature and trial files, source records, selection rules, retrieval code, task coding, and 16 joint-reader model files from this package. It adds a source residual score and matched early task calibration.
- usesdoi:10.18112/openneuro.ds004809.v2.2.0
The analyzed source is OpenNeuro ds004809 version 2.2.0; saved metadata and source records identify this version.
- doi:10.1038/s41467-017-02753-0
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesA fixed covariance correction increased mean recall forecast loss in 16 SEEG session pairs after five early lists at C=0.01
Its one-score test in the same panel supplied the task-calibration reference and method context for the present full-covariance test.
- usesShared SEEG weights lowered mean recall forecast loss by 0.00173 nats/word in 16 reused session pairs after five-list calibration at C=0.1
Used its uncertain local neural correction result to choose a test of brain weights learned from other patients. This study uses six channel-mean bands and one target intercept shift. The same patients were reused, so this is not an independent replication.