A fixed SEEG shape rule gave 15.6 percent stimulation pair accuracy after pulses and 4.5 percent before pulses in five single runs

We tested whether normalized SEEG waveform shape identifies an electrical stimulation pair on later pulses. We used public OpenNeuro ds004457 version 1.0.2. The sample had five patients, one run per patient, and eight fixed receiving contacts per patient. Each contact had 20 to 37 candidate pairs. The rule used the first five pulses to form signed unit templates. It classified the last five pulses without a parameter search. Mean accuracy was 15.58 percent from 50 to 250 ms after pulses and 4.50 percent before pulses. The equal-patient difference was 11.08 percentage points. Its descriptive 95 percent patient bootstrap interval was 7.84 to 13.99 points. Each patient had a positive difference. Candidates on one shaft gave 42.51 percent accuracy after pulses and 29.33 percent before pulses. The later window check also gave a positive difference in each patient. Most predictions were incorrect. The result applies to these stimulation blocks. Recording time, patient state, and stable artifacts can contribute. It does not establish neural connections or transfer to another session. The package preserves input data, all scores and predictions, code, source hashes, and reproduction instructions. A raw-file rebuild matched every input array. A second numeric run produced identical result files.

stereoelectroencephalography · electrical stimulation · waveform shape · classification · human intracranial recordings