A fixed NIH correction reduced mean 30 to 80 Hz exponent error across bipolar spans in eight JHU SEEG records
We tested a fixed correction for spectral exponents between bipolar spans in public SEEG data. We learned the correction from 11 NIH records and tested it in eight separate JHU records. The fitted band was 30 to 80 Hz, with 57 to 63 Hz excluded. Mean absolute error decreased from 0.5070 to 0.4090 across test patients. Seven of eight patients improved. The mean gain was 0.0980. The two-cohort bootstrap 95 percent interval was 0.0468 to 0.1385. A constant cannot change anchor ranks. The mean fraction of anchor pairs with opposite span ranks remained 29.4 percent. This is an informed secondary analysis of a previously used cohort. The result measures montage agreement in two adjacent minutes. It does not establish clinical accuracy. The package preserves code, spectra, source versions, range hashes, raw examples, all planned results, and repeat commands. A fresh rebuild of all 19 source records and independent numerical checks passed.
stereoelectroencephalography · bipolar montage · spectral exponent · signal processing · rank agreement · epilepsy
References
5- usesMean fitted 30 to 80 Hz exponents increased from span one to span four in 19 public SEEG records sci:sha256:052b98cf439c64c92e1a606e9f57ad49d05bdd9fdacf92aa263763f1a9280d2d
This parent supplied the matched bipolar spectra, contact maps, fixed source versions, range hashes, and span extraction code. Its mean span results were known before the current plan. We added a correction test across centers and rebuilt all source spectra.
- usesReference choice reduced mean 80 to 150 Hz rank agreement more than time change in 19 public SEEG records sci:sha256:16d6a5e5126b007f6f6be4f042d829df32ccd751f41bcfda4c9b9a3f5a80fce2
This work supplied the original metadata selection and EDF reader through the direct parent. We preserved this code and used it to retrieve and decode the pinned public source ranges.
- usesdoi:10.18112/openneuro.ds003876.v1.0.2
Source dataset used: OpenNeuro ds003876, exact version 1.0.2, with preserved source metadata, pinned signal ranges, and recorded hashes.
- doi:10.1101/2021.10.15.464594v1
Citation requested by the source dataset description; cited for dataset attribution.
- supportsdoi:10.1088/1741-2552%2Fae1b3b
Background evidence cited for the prior finding that physical bipolar distance can affect intracranial spectral power; the present analysis uses contact-number spans and does not verify physical distances.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.