Early half-rate fits did not establish a median-channel beta prediction gain over pre-stimulation in seven SEEG patients with about 43 Hz stimulation

We tested whether an early fit at half the stimulation pulse rate predicts later beta voltage in public SEEG records. The primary sample had 11 records from seven patients in OpenNeuro ds006519 version 1.0.0. We fitted sine and cosine coefficients in an early window and froze them for a separate later window. We compared the resulting beta error reduction with matched pre-stimulation windows. The median-channel difference was 1.93 percentage points. Its patient bootstrap 95% interval was -0.78 to 4.78 points. The two-sided sign-flip p value was 0.28125. This result did not establish a median-channel gain over the control. A fixed energy-weighted check gave 30.32% error reduction during stimulation and a control difference of 37.51 points. A descriptive check showed that a small channel group held much of the beta energy. The package also includes a 50 Hz comparison and four source metadata frequency conflicts. It preserves exact inputs for 31 records from 19 patients, four figures, source metadata and hashes, Python code, result tables, and a raw retrieval recipe. A separate source rebuild reproduced all numerical inputs and results exactly. The analysis cannot separate instrument artifact from driven neural voltage. It does not validate neural decoding or a live BCI.

stereo-EEG · electrical stimulation · beta voltage · half-rate prediction · signal processing · brain-computer interfaces