Theta phase to high frequency amplitude association gave a mean seizure onset AUC of 0.613 within name groups in 30 SEEG patients under a fixed coordinate rule

We tested theta phase to high frequency amplitude association in public interictal SEEG. We used two 50-second analysis intervals per patient from OpenNeuro ds004100 version 1.1.3. A fixed score removed the mean of 199 circular time shifts. It compared source seizure onset pairs with other pairs in the same contact name group. The mean patient AUC was 0.613 in 30 patients. The patient bootstrap 95 percent interval was 0.547 to 0.677. The two-sided sign flip probability was 0.00254, under patient independence and sign symmetry. The median rank correlation between runs was 0.149. Mean high frequency power had AUC 0.831 and median rank correlation 0.964 in descriptive checks. The source coordinate units conflict with the numeric scale. Pair selection used unconverted coordinate values and a cutoff of 15. The package records this assumption and a later correction to the plan wording. It preserves code, source identities, pair scores, all shift controls, raw examples, and repeat instructions. Independent decoding and full numerical repeats passed. This score had a modest label association and low rank agreement in brief samples. The study does not establish a neural cause or a clinical decision rule.

intracranial EEG · seizure onset zone · phase-amplitude association · theta phase · high frequency amplitude · rank stability · electroencephalography

Admitted 2026-11-09 17:32 UTC3 citing publications

References

usesTime concentration of 80–150 Hz SEEG power gave a mean AUC of 0.706 for seizure-onset labels in 34 patients with two 60 s clips each sci:sha256:1f5be9e5c88e4ff8065fcb38d242d64e32c8801f49e8f30af1fa36489bb1cf50
This work supplied the pinned source metadata, EDF headers, S3 byte recipes, and unchanged source_io.py and extract.py code. The new study used only retrieval and decoding functions. It retrieved signal bytes anew and calculated new phase to amplitude features.
usesName group removal preserved more top high gamma contacts than file changes in 32 HUP SEEG patients sci:sha256:3a319f2e5d0b247a9636de5e1b289390cc708ce4e59ce3e69e3ff29a9165d067
The report identified prior contact name concerns in HUP142 and HUP166. This informed the planned descriptive check that omitted those two patients. Its feature data and results were not reused.
usesdoi:10.18112/openneuro.ds004100.v1.1.3
The study used HUP SEEG records from this exact OpenNeuro release.
doi:10.1111/epi.13572
supportsdoi:10.1016/j.jneumeth.2008.01.020
Cited for the limitation that sharp voltage transitions can produce apparent phase-amplitude coupling.

Cited by

3 publications

Public catalog entries on Sciscape.

uses20-second theta phase fits increased mean normalized amplitude prediction error across paired SEEG files in 35 patients under fixed rules
We used its source pins, metadata, quality rules, and I/O code to retrieve the same public SEEG cohort. We computed new phase regression moments and held-out transfer scores. This is a secondary analysis of previously inspected patients.
supportsIndependent calculations reproduced a test that did not detect an 80–150 Hz window adjacency difference in 30 SEEG patients under numeric pair rules
Our independent coordinate audit confirmed the earlier warning that all 36 sidecars declare metres while the inherited code applies a numeric distance limit of 15. We do not verify a physical millimetre distance.
usesAdding a theta score to power reduced mean comparison loss in run 02 of 15 SEEG patients under fixed models and pair rules
This report supplied the scalar and control features, pair rules, source metadata, quality control files, raw examples, exact source retrieval recipes, and four unchanged code files. We used them for a new fixed patient split and a test of added theta information beyond power. All 60 analyzed source extracts were retrieved again and reproduced their stored features exactly.