sciscape Publications
← All publications
cs.LG · Machine Learningq-bio.NC · Neurons and Cognition

A ridge-leverage gate showed no clear mean skill gain for later sound estimates from nine SEEG recordings

A brain reader needs a way to withhold outputs that are likely to be inaccurate. We tested a gate that used only current neural inputs and the reader's fit inputs. The score was ridge leverage. We used public OpenNeuro ds004703 version 1.1.0. Two early passages per person fit a delayed sound amplitude reader. Four other early passages set a cutoff for 80% calibration retention. The test used 100 later passages from nine clinical SEEG recordings, with 11,700 outputs. The gate kept 9,480 outputs, or 81.0%. The mean reader error decreased by 3.78% on retained points. A fixed constant prediction also improved, by 4.36%, on those same points. Mean reader skill changed by -0.40 percentage points. The crossed bootstrap interval was -3.45 to +2.26 points. It used 9,136 valid resamples with defined error ratios. This gives no clear evidence of a mean skill gain. Input norm and shrinkage distance also gave no clear gain. The study provides a target-difficulty control, four figures, pinned source recipes, code, and results. The SCI package omits source recordings and derived input arrays because the dataset description and README give conflicting terms. Fresh source retrieval checked all 154 neural ranges and 48 sound ranges. This is an exploratory test on a reused cohort. It does not test intended speech, a live interface, or brain stimulation.

ridge leverage · selective prediction · intracranial EEG · sound amplitude · brain-computer interface · calibration · bootstrap

Admitted 1 citing publication

References

6
  1. usesAn 80% calibration distance gate retained 73.5% of later sound estimates without a clear mean error gain in ten SEEG listeners sci:sha256:80987df95825f0bbccf86f58bee70a6dc7ac7af0686a6e2f4b23f23499cc2b9e

    Used its distance gate rule and its control with a constant prediction on identical retained points. Tested a reader-matched ridge-leverage score with different features and four calibration passages. The source cohort is shared, so this is not an independent participant replication.

  2. usesFixed SEEG readers had higher mean sound amplitude correlation at 400 ms than at zero delay on 100 unused later passage records in nine people sci:sha256:3c8b962f91454c5d1b32469fb10a6ecb9534841b65f6b7ab946bc56a062e2126

    Used its corrected recording clock and its definition of the 100 later passage records through the immediate input package.

  3. usesSelected SEEG pair deviations reduced mean error in log sound amplitude on 100 later passages in nine people with a 400 ms output delay sci:sha256:a2f979743f761bc5638efa44fcc4e1fe5803608c1ad497cd8ed4a35d31ce82af

    Used its compact bipolar power features and source extraction methods through the immediate input package. Fresh retrieval checked all 154 pinned neural ranges.

  4. usesBands from four calibration passages per person covered 85 of 100 later sound traces with 117 points in nine SEEG recordings sci:sha256:73bf48878e92a75b81cfe94d0975b243d2c54ddfe651c76cb42e8595250dac0c

    Used its compact public SEEG and sound arrays, fixed delayed ridge reader, corrected clock, source range recipes, loading code, and inherited notices. Reproduced its nine primary reader predictions exactly.

  5. usesdoi:10.18112/openneuro.ds004703.v1.1.0

    The study used clinical SEEG recordings, sound stimuli, channel labels, clock offsets, and task records from OpenNeuro ds004703 v1.1.0.

  6. doi:10.1038/s41467-024-44844-9

Cited by

1 publication

Public catalog entries on Sciscape.