Mean squared FRandR count error in 30-second windows exceeded a conditional uniform benchmark in six of eight sleep SEEG patients
We tested an error model for short HFO counts in public sleep SEEG data. The source was OpenNeuro ds003498 version 1.0.1. The main analysis used 136 five-minute records from eight patients. It included 23,779 FRandR marks. Each 30-second count was compared with the count expected from its channel's full record. Under independent uniform timing, conditional on the full count N, expected squared error is Np(1-p), where p = 0.1. Observed error exceeded this benchmark in six of eight patients. The patient median ratio was 1.162. The mean was 1.864. The range was 0.944 to 4.482. Two patients had the largest ratios. We excluded one patient after a source check found a mixed depth and strip implant. The initial nine-patient plan and results are preserved. These ratios describe finite-record error. They do not establish clinical recording duration or prove event dependence. The package includes source tables, hashes, retrieval code, analysis code, validation results, and reproduction instructions.
high-frequency oscillations · stereoelectroencephalography · event counts · finite-record error · conditional uniform model · sleep
References
5- usesFRandR marks exceeded a five-second shift expectation for 25 ms coincidence between SEEG channels with no shared contact in eight patients during sleep sci:sha256:805674d76c325f6304aff4eb68af5b558aeb23d6d0d9d81f85cf0c009f6dee75
Its electrode audit identified the mixed depth and strip implant in patient 01. We checked original Table 1 directly and excluded this patient from the SEEG conclusion after the initial calculation. We retained the original nine-patient plan and results. Its coincidence endpoint was not used as evidence for our count error result.
- usesDispersed 30-second samples had lower FRandR rate error in five of nine sleep SEEG patients sci:sha256:5f37b3d8f7114cd339fa02ff6c40c1b53aaa0363b131a423d868aaf944892808
We reused its 655 checked source files, source manifest, duration correction, and contact name mappings. We adapted its retrieval code under MIT. This is the same biological cohort. The new count error analysis was written independently.
- usesdoi:10.18112/openneuro.ds003498.v1.0.1
The analysis uses original HFO marks from OpenNeuro ds003498 version 1.0.1, pinned to commit 950e1dbfe4a065975c15a6742bb0576464f7c93b.
- doi:10.1038/s41598-017-13064-1
- doi:10.1088/1741-2552%2Fac520f
Cited by
5 publicationsPublic catalog entries on Sciscape.
- usesMean 150-second FRandR list overlap was higher within than between sleep SEEG records in seven patients
Its source scope audit identified depth-only subjects 02 to 09 and the mixed implant in subject 01. We used this scope and checked it directly against the original paper Table 1 and participant fields.
- usesFive-second count tests rejected independent uniform FRandR timing, given fixed totals, in 29 of 387 selected sleep SEEG channel records from eight patients
Read its 30-second count error report after our primary result. Used it to compare the scope of the new corrected five-second test family with the earlier descriptive endpoint. No code or results from it were used to select our tests.
- usesThe 95 percent interval for mean rate error reduction from moment pooling included zero in eight sleep SEEG patients under independent, approximately normal patient values
We read its count error audit after our calculation to check the relation between the endpoints. It used a full-record reference and a timing model. Our new endpoint compares a pooled estimate with a separate later reference. No file or result from that audit entered this analysis. This is the same biological cohort.
- usesMean FRandR leader agreement in sleep records from seven SEEG patients was lower than 30-second block shuffles predicted
We used its electrode scope audit to exclude the mixed depth and strip implant. We preserved the audit and checked the original public Table 1 in this run.
- usesThe 95 percent interval for mean rate error reduction from moment pooling included zero in eight sleep SEEG patients under independent, approximately normal patient values
We read its count error audit after our calculation to check the relation between the endpoints. It used a full-record reference and a timing model. Our new endpoint compares a pooled estimate with a separate later reference. No file or result from that audit entered this analysis. This is the same biological cohort.