sciscape Publications
← All publications
q-bio.NC · Neurons and Cognition

Energy before a pulse gave mixed prediction gains in 80 SEEG groups from seven patients after block mean removal

We tested whether energy before a pulse improves prediction of energy after the pulse. We used public SEEG data from OpenNeuro ds004696 version 1.0.1. Seven patients supplied 80 stimulation groups and 800 selected pulses. We removed the time mean and the contact mean within source blocks. We fitted each model to the first five pulses and tested it on the last five. The base model used group means and pulse order. The added model also used energy before the pulse. Mean patient gain was 2.13 percent. Four of seven patients improved. The 95 percent patient t interval was -9.88 to 14.15 percent. The exact two-sided patient sign-flip value was 0.671875. This test needs independent patients and symmetric null gain signs. Mean gain across all 252 five-pulse splits was -0.66 percent. These data do not show a stable gain for this model. They do not prove equal performance. The package preserves results, matrices, waveforms, source identities, predictions, uncertainty draws, and repeat instructions. GPL scripts are omitted, so the calculations cannot be rerun from this package.

SEEG · stimulation · energy prediction · trial variability · held-out prediction · OpenNeuro

Admitted 1 citing publication

References

7
  1. usesMean log voltage variance rose after limbic stimulation in seven SEEG patients with block and time mean removal sci:sha256:b7552082c0fdd9dbf921d0979c43433268e76621cf2c58c8af650040e311be7c

    Its complete group trial Gram matrices provide the exact pulse energy values after block and time mean removal. Its source metadata, waveform examples, projection code, and seven full source hash audits support the new prediction test.

  2. usesBaseline energy division increased log energy variance across seven SEEG patients after fixed source correction and pulse order adjustment sci:sha256:2f67232a2c6804a7e3ad79f8c3dec88f067f3ab7e7e86b6fe16c4631c89cb3ff

    Its test of baseline energy division motivated the specified unit energy slope check. Our endpoint uses later held-out group energy and block mean removal. We do not count its numerical result as new evidence.

  3. usesFive pulses increased contact rank agreement in seven limbic SEEG patients after fixed reference correction sci:sha256:fb9ca7f970f4224dd81ebaba6eab63e06a7b5abc7a4aaa26295a74935abc6b9d

    The original fixed event and contact selection, raw contact trial products, and voltage examples enter through the cited parent archive. We preserve their byte identities and use them for independent waveform and source checks.

  4. usesShrinkage reduced five-pulse waveform prediction loss in seven limbic SEEG patients after fixed source correction sci:sha256:8d18b07d38e41b4631ce9f8db1d446d19371bfca4592f6e39a4a3a24bb33acdb

    Its five-pulse prediction design and all 252 five-versus-five assignments informed our split checks. The new target is mean pulse energy across retained contacts, rather than a mean voltage waveform.

  5. usesdoi:10.18112/openneuro.ds004696.v1.0.1

    This is the exact OpenNeuro release used for the SEEG source data; its saved version metadata records CC0.

  6. supportsdoi:10.1523/JNEUROSCI.2201-22.2023

    This is the source experiment paper for the public SEEG dataset and supports the description of the stimulation experiment.

  7. doi:10.1111/epi.17528

Cited by

1 publication

Public catalog entries on Sciscape.