The patient resample interval for added encoding high gamma recall prediction gain included zero after fixed spectral controls in SEEG data from seven Utah patients

We tested added encoding high gamma value after fixed controls for beta power and the spectrum outside high gamma. We used OpenNeuro ds006136 version 1.0.1, with seven Utah patients, 85 released channels, and 846 later scored trials. The recording metadata identify SEEG. The paper reports depth implants, but individual electrode types remain unresolved. Added high gamma reduced equal patient mean squared error by 0.002342, or 0.445 percent of control error. The 95 percent patient resample interval was -0.000693 to 0.005490. Four patients improved and three worsened. The exact sign p value was 0.234375 under independent patient gains and a symmetric null. The constant model had lower mean error than the full model. The added high gamma gain remains uncertain in this fixed test. The package contains new spectra, features, all predictions, source pins, code, and reproduction instructions. Independent source and solver checks passed. A complete repeat matched all numerical outputs. The study does not establish a zero effect, a causal mechanism, or prediction on new patients.

working memory · stereoelectroencephalography · high gamma · spectral controls · recall prediction