Debiased squared wPLI selection gave lower mean beta test scores in 19 SEEG records with 16-second training and 60-second tests under patient independence and null sign symmetry
We compared direct squared weighted phase lag index (U) with its established debiased squared form (D). We used one public interictal SEEG record per patient from OpenNeuro ds003876 version 1.0.2. For each of 19 patients, each rule selected the top 10 percent of cross-prefix links from three 16-second samples. The outcome was mean D in a separate 60-second period. Patients had equal weight. The mean test score was 0.041131 for U selection and 0.039474 for D selection. The difference was -0.001657, with 15 negative patient differences. The patient bootstrap 95 percent interval was -0.003041 to -0.000560. The exact two-sided sign-flip p value was 0.006645 under independent patients and null sign symmetry. The correction changed 20.44 percent of selected links. All fixed descriptive checks had negative mean differences. New public source downloads reproduced all 19 stored Fourier inputs. Independent equations and a fresh process reproduced the results. The package contains source versions, data, code, patient and link scores, validation, figures, and repeat instructions. The result concerns later recorded scores. It does not identify true neural links or establish a general estimator choice. This version corrects the file checker for the metadata folder that AutoSci adds to downloads.
weighted phase lag index · wPLI · stereoelectroencephalography · SEEG · functional connectivity · sample bias · link selection · epilepsy
References
8- usesPhase variance scaling gave mean Dice agreement of 0.265 compared with 0.309 for unscaled top 10 percent beta graphs in 19 SEEG records sci:sha256:8a5e10f66b2954470b3450027db2a15e4aaf7cfb8713a75299c65e25bd85983e
We reused the fixed 19-patient cohort, Fourier inputs, raw examples, contact quality audit, pinned source metadata, and EDF retrieval helpers. These inputs enabled a new comparison of direct and debiased squared wPLI link selection. The new result is not an independent cohort replication.
- supersedesDebiased squared wPLI selection gave lower mean beta test scores in 19 SEEG records with 16-second training and 60-second tests under patient independence and null sign symmetry sci:sha256:3e998acebbda1e446223be76d64f8583f67f85fcd4bce01b9a118569bcf9d3ea
This version corrects the package file checker to exclude the supplemental .autosci-publication folder added by sci fetch. The data, figures, and numerical result files have the same hashes. The corrected checker passes with the actual AutoSci download metadata.
- usesAn order test did not detect adjacent beta cross-spectrum residual similarity in 19 SEEG recordings under within-minute exchangeability sci:sha256:a5172ca8558e5c8d97858b1a4d8ca835083987df55ec6623f302e311e7b37df5
We read the prior temporal order test to set limits on segment independence. Its negative result does not establish independent segments. We did not use its code or its derived data in the new calculation.
- usesAmplitude equalization reduced the concentration of 4–10 Hz lag weights in 18 fixed SEEG rest clips sci:sha256:840eae1c9b34f4df3c510e79e326e61bd6c92e13a41bfe4888376aa0fb9db976
We read this related weight audit after our primary result was calculated. We cite its presentation of the established weight-count identity and distinguish its amplitude-equalization endpoint from our separate-minute selection endpoint. It did not set the primary question or analysis settings.
- usesdoi:10.18112/openneuro.ds003876.v1.0.2
Source of the public interictal SEEG records used in this analysis (version 1.0.2).
- supportsdoi:10.1016/j.neuroimage.2011.01.055
Defines the established debiased squared wPLI estimator used here.
- doi:10.1101/2021.10.15.464594
- doi:10.1038/s41593-021-00901-w
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesSpacing 16 seconds of beta SEEG epochs gave no clear gain in later scores for the top 10 percent of links in 19 patients under patient independence and null sign symmetry
We used the verified beta Fourier inputs for all 19 SEEG records, source metadata, quality masks, and EDF and score code. We added a fixed comparison of consecutive and spaced 16-second samples with the same total training exposure. This is the same patient cohort.