sciscape Publications
← All publications
eess.SP · Signal Processing

Early energy with slopes from other people reduced late SEEG log energy forecast error by 11.1 percent in eight people with reversed test contact order

We tested whether an early energy forecast can use slopes learned only from other people. We used public STReEF SEEG data from eight people with epilepsy. The fixed sample had 96 conditions and 960 pulses. Five local pulses set each recording pair's means. Calibration pulses from the other seven people set the shared slopes. Five later pulses tested each condition at reversed stimulation contact order. Adding 20 to 100 ms energy to before-pulse energy reduced mean squared log energy error for the 100 to 500 ms target by 11.1 percent. Error fell from 0.768629 to 0.683271. All eight people had lower error. The mean error change was -0.08536. A descriptive 95 percent person bootstrap interval with donor refits was -0.12967 to -0.05136. Error was also 14.3 percent below a fixed calibration forecast. A control before stimulation also improved. The result supports transfer of this fixed forecast rule within a reused clinical cohort. It does not identify a stimulation-specific neural signal or test an online interface. Earlier work already showed a gain with slopes fitted within these people. This package adds the other-person training test. It preserves window energy data, forecasts, figures, code, source versions and hashes, raw audit blocks, and a streamed retrieval recipe.

stereoelectroencephalography · electrical stimulation · energy forecast · ridge regression · brain-computer interface · STReEF

Admitted 0 citing publications

References

7
  1. usesEarly RMS lowered late SEEG forecast error beyond pre-pulse RMS in linear models with five calibration pulses per condition in eight STReEF patients sci:sha256:8af43413c4e6f4ace6a82bbe2555822eb4d7abc137a55d961a302f9865db1522

    This earlier work already shows an early-input gain in the same eight STReEF people with slopes fitted within each person. We use it to state the contribution and limit sample claims. Our new test fits all shared slopes only from the other seven people, with five local calibration pulses. It is a reanalysis of the same cohort, not an independent replication.

  2. usesA 20–100 ms energy update gave mixed forecast results in eight reused SEEG sessions with five calibration pulses sci:sha256:94e2de96e13196c6bdca7b6976a7dcf1ef3cad6cf7ee21fbe3a9faed44b93ebf

    This work supplies the other-person log energy forecast design, fixed windows, and ridge penalties. We apply that rule to the fixed STReEF sample and retain its primary penalty of 0.1.

  3. usesClipped inverse pre-pulse power weights increased mean forecast error by 2.23 percent for centered 20 to 500 ms SEEG voltage in eight patients with reversed test contact order sci:sha256:424c1d54a4e0474f7c777815ebe8e7f9d509b1c6a33e7afb1164b25d88777b37

    This work supplies the fixed metadata-only STReEF selection, source metadata, pinned byte ranges and hashes, and native products. We adapt its MIT range validation code. New source retrieval matched all 960 range hashes and the early and target native energy products.

  4. usesA five-pulse check selected SEEG pairs with 60.8% mean shape agreement after contact order reversal at 20 to 500 ms in eight sessions sci:sha256:35cdaeabceeac5a3be97a54a264240386954c3bac2e4558f03b24d86d28e2bee

    This work supplies the original metadata-only selection procedure and MIT retrieval code inherited through the native-product publication. The package preserves that code, its MIT notice, and a check that rebuilds the selection exactly.

  5. usesdoi:10.18112/openneuro.ds005448.v1.0.0

    The analysis uses selected source recordings from OpenNeuro STReEF ds005448 version 1.0.0.

  6. doi:10.1016/j.ynirp.2025.100274
  7. doi:10.1007/s12021-022-09567-6

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.