Mean 6 dB rise detection increased after marked seizure onset in paired 10 s periods from 26 selected SEEG patients

We tested a fixed spectral rise rule before and after marked seizure onset. We used public stereoelectroencephalography (SEEG) data from OpenNeuro ds004100, version 1.1.3. The study used two seizures from each of 26 selected patients. It included 2,741 local bipolar pairs in 332 contact name groups. Fast power was 20 to 50 Hz. Slow power was 2 to 12 Hz. A detection needed a 6 dB ratio rise in two consecutive one-second blocks. Each seizure had its own earlier baseline. We compared the last 10 seconds before source onset with the first 10 seconds after it. We gave equal weight to groups, seizures, and patients at each level. The mean detection fraction increased from 10.90% to 28.70%. The paired increase was 17.80 percentage points. The 95% patient bootstrap interval was 10.97 to 25.75 points. The increase was positive in 22 patients and negative in four. All seven fixed descriptive checks had positive mean increases. This result shows time separation for one rule in one selected cohort. Pre-onset detections cannot be treated as clinical false alarms. The package contains source records, code, compact signal features, results, figures, and repeat instructions.

seizure onset · stereoelectroencephalography · SEEG · spectral ratio · signal detection · epilepsy