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

Training with nine window shifts raised mean site accuracy under uniform −20 to 20 ms shifts but reduced accuracy at zero in five SEEG sessions

We tested training with shifted response windows in five public SEEG sessions from OpenNeuro ds004457 version 1.0.2. The test used 183 stimulation sites. Each site supplied four training pulses and four separate test pulses. Nine test shifts from -20 to 20 ms in 5 ms steps had equal weight. Training with the same nine shifts raised mean site accuracy from 17.79% to 21.79%. All five patients had a gain. At zero shift, mean accuracy fell from 32.93% to 25.27% in every patient. The primary two-sided patient sign-flip p value was 0.0625. A later, exploratory component swap showed a smaller wide-shift gain from changing site means alone. This reused offline split has a known site-block schedule limit. The results do not establish a population gain or live brain-interface performance. The package preserves code, input bins, every decision, fitted models, three figures, source hashes, and instructions. All public raw source files passed hashes. A raw rebuild and complete numerical replays matched the saved inputs and results.

stimulation responses · SEEG · response window shifts · site classification · classifier robustness

Admitted 1 citing publication

References

7
  1. usesA joint SEEG shape decoder did not show a clear mean gain over one selected pair in four patient sessions sci:sha256:81f77d4f52365d5b14fed7d3d6db7b430a2cf478b9a6a6062540c4e90bbd8e7c

    This preparation ancestor supplied the contact targets and raw MEF extraction method through the direct input publication. We kept those choices and rebuilt the public source arrays.

  2. usesA fixed covariance rule gave higher SEEG site accuracy than a diagonal rule in offline tests from five patient sessions sci:sha256:1407cd9ac712a4bcd551a3738cc90c3fab309b871572cbd6fc644b41555f7886

    We used its fixed half-diagonal covariance distance rule. Our window is 40 to 120 ms, and our residual moment gives each physical pulse equal total weight across shifted views.

  3. usesPulse time alone gave 74.8% mean site accuracy in five offline SEEG tests with shared calibration and test blocks sci:sha256:07f3c9173d8e02cafb724475eb4d7d43bb6e6bf6e042867f73744b4ea8895b75

    Its voltage-free time control and site-block audit limit interpretation of our inherited offline split. We cite its measured schedule result and do not claim that site classification isolates biological site information.

  4. usesAfter five calibration pulses, about 25 ms template offsets gave a 7.48% mean patient error increase in 100–500 ms SEEG forecasts across eight sessions sci:sha256:949831979a0945dfe30f29a37ca4d255ab1dab3c525ac796fa997f1a4a3ed170

    Its offset test in late voltage forecasts motivated the window alignment question. We tested site classification and training with shifted windows in a different public dataset.

  5. usesRemoving missing features raised mean site accuracy for 20 to 100 ms responses under known simulated contact loss in five SEEG sessions sci:sha256:ee49903b43a054e26fbb0d85370a8cc3bbeb395ca7baf6c0069fcd7329eac507

    This publication supplied the binned voltage inputs, fixed contacts, site eligibility, pulse split, source metadata, and retrieval code. We used these files for a new window-shift training comparison.

  6. usesdoi:10.18112/openneuro.ds004457.v1.0.2

    Source dataset for the five SEEG sessions and derived voltage bins.

  7. doi:10.1523/JNEUROSCI.1325-22.2023

Cited by

1 publication

Public catalog entries on Sciscape.