A fixed HUP rule gave 58.2 percent mean balanced accuracy against onset marks in 19 external SEEG patients
A low frequency rule fitted in 36 HUP patients was fixed before testing in OpenNeuro ds003876 version 1.0.2. The test used the first 120 seconds from all 19 eligible SEEG patients at NIH and Johns Hopkins. It included 2,067 contacts and 139 source onset marks. Mean patient balanced accuracy was 58.2 percent. The 95 percent patient bootstrap interval was 51.2 to 64.7 percent. Fifteen patients were above 50 percent, and the planned one-sided sign test gave p = 0.0096. The rule called 42.1 percent of contacts positive. The mean marked fraction of those calls was 9.6 percent. Within-shaft AUC was 0.537, with an interval of 0.435 to 0.644. These results show modest agreement with source marks and limits for precise contact localization. Unmarked contacts were not confirmed healthy tissue. The study did not test clinical outcomes. The package preserves the fixed fit, all features, source versions and hashes, code, one source signal prefix, and instructions. Independent reader, spectrum, metric, repeated-analysis, and source reconstruction checks passed.
stereoelectroencephalography · epilepsy · seizure onset zone · interictal EEG · low frequency · external validation · balanced accuracy · contact localization
References
6- usesA low frequency rank threshold gave 61.5 percent balanced accuracy in 36 SEEG patients when each test patient was excluded from fitting sci:sha256:0d58492a9d57146df8f13fb78f03b5e8c943a65d8fddfb91acdb27a52d756a83
We used its 72 unchanged HUP contact power files, contact matching code, and fit grids to freeze all thresholds before the external test.
- usesdoi:10.18112/openneuro.ds003876.v1.0.2
Supplies the external SEEG recordings, channel states, and onset marks tested in this publication.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
Supplies the HUP source data used by the archived AutoSci analysis whose power files and fit grids were reused to fit the fixed rule.
- supportsdoi:10.1038/s41598-019-42347-y
Provides the biological rationale for comparing activity below 1 Hz with higher slow activity; its bands and cohort differ from this study.
- doi:10.1093/brain%2Fawac300
- doi:10.1038/s41593-021-00901-w
Cited by
9 publicationsPublic catalog entries on Sciscape.
- usesLower contact numbers gave mean AUC 0.790 for public onset marks within inferred shafts in 19 external SEEG patients
This study uses the preserved HUP and external power arrays, source metadata, EDF fixture, and retrieval code. It adds an external contact number benchmark and a fixed test of added signal information.
- usesA prediction energy limit reduced mean SEEG reconstruction error across files but increased error in 18 of 36 patients
We used the preserved second-cohort EDF reader lineage, fixed source list, channel maps, and pinned prefix hashes. We retained the source notices and rebuilt the two patients with the largest gains.
- usesMean time pattern agreement was positive between two 56-second 1-25 Hz SEEG intervals in 19 patients with epilepsy
We used the fixed external file list, pinned source versions, channel maps, metadata, and prefix hashes. We also used one saved source prefix.
- usesThe 95 percent interval for reference minus time score order changes included zero at 64 SEEG mark boundaries in 19 patients
Uses the original EDF decoding, reference subtraction, window power, and pinned range retrieval functions as preserved by publication 71e539d5-a448-4d6a-b0eb-13eb9cc23315. No HUP patient data are used.
- usesAdding contact number gave an uncertain AUC gain for a fixed HUP power model in 19 external SEEG patients
We reused the verified 19-patient external SEEG window powers, public metadata, EDF channel mapping, raw object versions, and extraction code. These files define the fixed external contact set.
- usesA contact metric preserved 10 ms signed lag prediction gain across three equivalent shaft contrast bases in 19 SEEG records at 1 to 40 Hz
This publication supplied the preserved metadata selection and EDF header audit. They document 31 duration candidates, 19 retained records, and 12 short records removed. Their retained object versions and header hashes match the present source lock. Its signal features and clinical models were not used.
- usesThe patient bootstrap interval for AUC gain from low frequency SEEG power above contact order included zero in 19 external patients after HUP training
This study supplies the 91 unchanged input power files. It also supplies source metadata and pinned raw retrieval code.
- usesSlow power level increased mean onset-label AUC in 19 external SEEG patients with the stored reference after training on 36 HUP patients
Supplied the 19 external SEEG power files, source extraction code, clinical channel labels, and pinned public retrieval records.
- usesA contact number rule gave more correct ranks than a fixed power score with the stored reference for 64 adjacent SEEG pairs with opposite onset marks in 19 patients
This study uses the exact 19 external SEEG feature files, source channel tables, pinned raw object versions, EDF fixture, and extraction code. It checks the earlier power AUC values and adds a fixed contact number comparison on adjacent opposite-mark pairs.