Under an acquisition clock assumption, a response model showed no clear modality transfer gain over prestimulus power in six SEEG recordings

We tested transfer between visual and auditory number formats in six public SEEG recordings. The source was OpenNeuro ds005415, version 1.0.0. The analysis assumes that acquisition samples use the public EDF time origin. Models trained on digits and spoken numbers tested on dots and beeps. We also tested the reverse direction. Mean AUC was 0.639 for response changes and 0.604 for prestimulus power. The mean paired gain was 0.035. Its 95 percent patient bootstrap interval was -0.103 to 0.244. The exact sign-flip test gave p = 0.9375. Both models exceeded shuffled-label benchmarks. The response model did not show a clear gain. This result does not prove equal performance. The package preserves source features, raw examples, code, scores, model fits, label shuffles, and source range recipes.

stereo-electroencephalography · modality transfer · prestimulus power · number formats · ridge regression · area under the curve · event timing · patient bootstrap