Recall feedback raised observed mean Brier error by 0.00082 for fixed high gamma models in 18 SEEG sessions
We tested recall prediction in 18 public SEEG sessions from OpenNeuro ds005059, release 1.0.6. Early lists supplied fixed high gamma models. Before each later list, we changed only the intercept. The update used outcomes from completed lists. We scored 930 later trials. Mean Brier error rose from 0.191525 to 0.192342. Six patients improved. The mean gain was -0.000817. Its 95 percent patient bootstrap interval was [-0.002502, 0.001010]. The updated neural model gave a gain of 0.001125 over an updated recall-rate control. Its interval was [-0.002052, 0.004163]. Both Holm-adjusted p-values were 0.780197. A separate diagnostic showed that pooled AUC can fall below 0.5 with correctly timed feedback and independent labels. Exact enumeration of two six-trial lists gave mean AUC 0.418144. This diagnostic was added after the primary results. The reused cohort does not provide independent confirmation. The offline replay assumes that scores are available after list completion and the last recorded response. The package preserves code, features, source records, predictions, checks, figures, and reproduction instructions.
stereoelectroencephalography · memory · recall prediction · Brier score · intercept calibration · high gamma · sequential prediction
References
4- usesHigh gamma models selected on early lists gave Brier gains of 0.0123 over task models and 0.0010 over recall rates in 18 SEEG sessions sci:sha256:8a18030ad75dd073bf74c34f236c6eae3168d0b377c65a97ba667d147b6ae016
This study uses penalties selected only within early lists and the fixed neural and task model fits. It rebuilds these choices and fits. The new update changes only intercepts after completed recall lists.
- usesA fixed high gamma model gave no clear recall prediction gain on later lists in 18 public SEEG sessions sci:sha256:8993bf239839d217822ccc5ea36a6591d7d751e2641e08eb2b6d107f2505b38a
This study uses the 18-patient SEEG trial features, labels, source records, range retrieval recipe, and logistic model solver. It preserves the original package and checks its file hashes.
- usesdoi:10.18112/openneuro.ds005059.v1.0.6
The analysis uses source data from OpenNeuro ds005059, release 1.0.6.
- doi:10.1016/j.neuroimage.2015.03.077
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.