A scale from other patients gave mean Brier error 0.187 in eight offline SEEG tests
We tested a confidence scale for eight fixed SEEG decoders with public OpenNeuro ds004473 version 1.0.2. Each scale used the middle blocks of seven other patients, with equal patient weight. The target decoder still used earlier labels from the target patient. Testing used 674 later trials, or 1,348 selected windows. Mean Brier error fell from 0.193261 to 0.186883. The paired mean change was -0.006377. A patient bootstrap refitted donor scales in 20,000 joint draws. The 95 percent interval was -0.015029 to 0.001035. Five of eight patients improved. A fixed temperature of 2 gave mean error 0.186454. This test did not establish a benefit from the learned scale. Rest-control Brier error fell from 0.302752 to 0.250520, while accuracy stayed 49.65 percent. Positive scales preserved every class decision and the score order. The package provides code, compact inputs, all scores, three figures, source hashes, and repeat instructions. Fresh raw checks passed for 16 source ranges. Independent fits and a repeat run passed. The result applies to selected windows in eight reused epilepsy records. It does not establish live command readout or brain writing.
stereo-EEG · confidence calibration · temperature scaling · Brier score · brain-computer interface · patient bootstrap · probability forecasting
References
5- usesWith eight pairs, omitting 70–110 Hz power reduced mean balanced accuracy by 5.6 points in eight offline SEEG tests sci:sha256:e3688bef3ed8761381d86408e300703e855ca508766aef3507db1ef78561ef10
This work produced the same eight-patient SEEG feature data used by the immediate parent. We preserved the inputs and checked selected features against fresh public source ranges.
- usesA rest threshold set for 5 percent errors gave 3.9 percent false detections and 26.5 percent sensitivity in eight later-trial SEEG tests sci:sha256:2100f971aed9c12a7d4be6af5c0b8c155e730ec5176d7b8c6dd80fcfd083230d
This work supplied the raw voltage retrieval and feature rebuild recipe used by the feature product. We preserved that recipe, source object versions, and range hashes.
- usesEarly confidence scaling decreased mean Brier error from 0.193 to 0.185 in eight offline SEEG patient tests sci:sha256:f3d79f04937a343945e7a53f5b54b00a0284174fca051959a818f1c79d41a7d1
We used its compact inputs, source records, fixed decoder choices, and chronological split. We read its target-only calibration results before this exploratory test. We tested a new scale that uses only other patients.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analyzed records are from OpenNeuro ds004473 v1.0.2, identified in the publication and pinned source records.
- usesdoi:10.1088/1741-2552%2Facae0a
This paper describes the source SEEG recordings and task used for the analyzed cohort.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesCorrection with unlabeled rate estimates gave higher Brier error than known rate correction in eight SEEG tests at a simulated 10 percent task rate
We use its positive score scale procedure and fixed bounds of 0.05 to 20. Our calibration block and fitted models are new.
- usesScores reduced mean Brier error with eight adaptation labels in eight offline SEEG tests at a simulated 10% task rate
The score scale method and bounds came from this calibration study through the direct parent.
- usesScores reduced mean Brier error with eight adaptation labels in eight offline SEEG tests at a simulated 10% task rate
The score scale method and bounds came from this calibration study through the direct parent.