Two components showed 0.35 percentage points more ordering gain after stimulation in eight patients, with an interval spanning zero
In eight patients with stereo-electroencephalography (SEEG), two principal components fit to five early pulses captured 29.83% of later residual variation after a pulse and 27.48% before a pulse. An exact reference for two uniform random directions in the same rank-four learned space captured 20.12% and 18.12%, respectively. Component ordering gain was therefore 9.71 points after and 9.36 points before stimulation, a paired difference of 0.35 points (95% patient bootstrap interval: -0.67 to 1.44; four of eight patients positive; sign-flip reference value 0.555). These results do not establish an extra ordering gain after stimulation or equality between windows. The interval allows small positive or negative differences. The reference controls for the learned space, not all brain or measurement differences. The analysis used 886 readout pairs at 86 sites in the same eight patients represented in OpenNeuro ds004696 v1.0.1; it is not an independent patient replication. The exact random-space reference separates transfer of the learned span from variance ordering within it. The work is exploratory, uses processed public waveforms, and does not test an online brain interface or identify a biological mechanism.
stereo-electroencephalography · SEEG · principal component analysis · random subspaces · electrical stimulation · limbic system · brain-computer interfaces · signal compression · OpenNeuro
References
4- Two components from five early pulses captured 29.8 percent of later SEEG variation at 20 to 500 ms after removal of the early mean direction in eight patients sci:sha256:40710d623f4d60ce0573eff904e924868aabf89d82baccd4f1ea7b426e43eded
[Legacy relation: reproduces] The new calculation reproduces 3,544 earlier rank-two and rank-four scores from the same archived inputs. The largest absolute difference is 3.33e-16. This checks the arithmetic. It is not a new patient replication.
- usesTwo components from five early pulses captured 29.8 percent of later SEEG variation at 20 to 500 ms after removal of the early mean direction in eight patients sci:sha256:40710d623f4d60ce0573eff904e924868aabf89d82baccd4f1ea7b426e43eded
This publication supplies the all-pulse product matrices, contact and event maps, source metadata, fixed waveform example, retrieval code, and inherited source checks. Its aggregate component scores motivated the exact random-space reference.
- usesdoi:10.18112/openneuro.ds004696.v1.0.1
The analysis uses processed waveform files and preserved event and channel labels from OpenNeuro ds004696 version 1.0.1.
- doi:10.1523/JNEUROSCI.2201-22.2023
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesExploratory held-out two-component PCA checks gave 80.7% mean rejection of digitally doubled SEEG waveforms at 20–500 ms after five fit pulses in eight patients
This earlier component study in the same cohort informed the choice of a two-component waveform model. Our residual checks do not add independent patients.