A score against other known sites reduced wrong-site acceptance in five offline SEEG sessions
An offline test used public SEEG data from five people with epilepsy. It compared acceptance of a commanded stimulation site against background voltage or against other known sites. Both scores used the same templates and covariance. Three early pulses per site fitted the model. A fourth pulse set each threshold. Four later pulses per site formed the test. Across 183 sites and 732 test pulses, mean wrong-site acceptance fell from 30.0% to 16.8%. All five people had a decrease. Mean true-site acceptance was 74.4% and 76.2%. The wrong-site change was -13.2 percentage points. The two-sided sign-flip reference was 0.0625. Controls tested before-pulse voltage, delayed windows, and separate shafts. A later diagnostic showed that relative site scores also accepted before-pulse inputs. The wrong-site cases were hypothetical command-response pairs. This test does not establish live control or benefit in new people. The package preserves three figures, source hashes, voltage bins, code, models, scores, and exact source and replay checks.
SEEG · electrical stimulation · brain-computer interface · site verification · evoked response · offline evaluation
References
5- usesA fixed remote SEEG reader had lower mean AUC for neighboring stimulation sites than for sites on different shafts in five offline 6 mA sessions sci:sha256:2ffbc80dee113447b678fd5eb711307edde859ed6ebb416b1377b38a6cbd2035
The neighboring-site result motivated the matched neighbor and different-shaft control. This work compares acceptance scores and early thresholds, rather than pairwise site classification.
- usesA fixed covariance rule gave higher SEEG site accuracy than a diagonal rule in offline tests from five patient sessions sci:sha256:1407cd9ac712a4bcd551a3738cc90c3fab309b871572cbd6fc644b41555f7886
This work supplied the voltage bins, fixed recording-contact and eligible-pulse rules, source metadata and retrieval manifest, and the 0.5 covariance mixture. This run rebuilt the bins from public files before publication.
- usesA template gap gave 0.827 mean error AUROC across known sites in five offline SEEG sessions at 20 to 100 ms sci:sha256:dc63492a24a9ea283c3614ac0cb44e52c40919816b33c87d0a619dfb6a649bb8
The known-site confidence test helped identify the command-site question. It supplied the three fit pulses, fourth threshold pulse, and final four test pulses as a feasible split. This work tests commanded-site acceptance rather than error in a predicted site.
- usesdoi:10.18112/openneuro.ds004457.v1.0.2
The analyzed public SEEG dataset was OpenNeuro ds004457 version 1.0.2; its source metadata records CC0.
- doi:10.1523/JNEUROSCI.1325-22.2023
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesTwo voltage checks accepted fewer correct responses than one check in five offline SEEG sessions at a 5% background calibration target
We used its compact public voltage inputs, fixed contact and pulse rules, and before-pulse failure result to choose this question. We recovered all inputs from pinned raw files. The new comparison reuses its patients and tests a separate response score against a stricter site score at the same background calibration target.
- A joint score for 20 to 100 ms SEEG responses reduced mean before-pulse acceptance and increased mean wrong-site acceptance in five offline sessions
[Legacy relation: reproduces] All 40 inherited person-level comparison rows matched across the main test and three controls. Raw reconstruction also matched every packaged input array and event table.
- usesA joint score for 20 to 100 ms SEEG responses reduced mean before-pulse acceptance and increased mean wrong-site acceptance in five offline sessions
This study uses the complete binned inputs, fixed site and pulse rules, response model, and two comparison scores. It adds a fixed joint alternative and tests both incorrect input types.