Eight donor shapes reduced five-pulse SEEG forecast error versus eight cosines at 86 selected sites in eight patients; full waveforms had less error
We tested whether response shapes from other patients improve a compact forecast of voltage after brain stimulation. The source was public SEEG dataset ds004696 version 1.0.1. We used 86 fixed selected readouts in eight patients. Each site used five early pulses to fit and five late pulses to test. The target patient was excluded from the learned shape basis. Eight learned shapes reduced mean patient normalized squared error by 11.30% versus eight cosine shapes. All eight patients had less error in this comparison. The descriptive fixed-basis resample interval was 6.46–17.50%. The same learned forecast had 7.64% more error than a full waveform forecast. This is an offline result in reused epilepsy recordings. It does not establish performance in a new cohort, live brain control, or a reduction in total model storage. The package preserves pulse windows, provenance, fitted shapes, compact coefficients, forecasts, figures, code, and source retrieval instructions.
stereoelectroencephalography · electrical stimulation · evoked potentials · signal forecasting · patient-excluded validation · low-rank models · response shapes
References
8- usesA fixed pre-pulse RMS rule increased mean SEEG interval scores after five fit pulses at 86 selected sites sci:sha256:500ee3502e00248d440a6ba1a465c4f3e91b148e8390a72ae1979088798e1843
We preserved its earlier full-source checks for all eight patients and its source event-order correction. These older checks are marked as inherited and are separate from this run's two fresh voltage checks.
- usesTwenty cosine coefficients increased mean 20–500 ms SEEG forecast error versus full scaled templates in eight patient sessions with five calibration pulses sci:sha256:f0178c1dc521dcf46916de948b41e3ce68a95c5741364315354453d40f62b230
This study supplied the fixed cosine forecast and full waveform comparison. We used those model forms at rank eight and tested a basis learned from other patients. Our sample has 86 selected readouts, rather than its 886 eligible readouts.
- usesFive calibration pulses gave mean later signed template AUC of 0.939 compared with 0.859 for one pulse in eight SEEG sessions sci:sha256:1d2aa103275ad7400befec4c899740ede8f2eb59a09897bdc3f4d4e190e54418
We retained its first-five-pulse response-energy readout choices and preserved its selection code. No readout was chosen from the new forecast errors.
- usesA fixed background power rule increased 20 to 500 ms SEEG size forecast error after five calibration pulses in eight sessions sci:sha256:b8e5ae61a5b00cfa06f9294e3694b2df810dda59e3ec09aa69ad2cbe63b674c4
We adapted its hashed byte-range download and source extraction recipes. Fresh full-source checks reproduced the selected voltages and event maps for patients 05 and 08.
- usesFive fit pulses and three calibration pulses gave 75.6% mean patient and 70.3% pooled whole response coverage at 86 selected SEEG sites sci:sha256:c3a61e7caa74d1cbc66131912586aa43ff1af13523ba1b58f0af0b4d0aa31373
This publication supplied the selected voltage windows, pulse maps, source metadata, and fixed choices for 86 readouts. We used the first five pulses to fit and the final five pulses to test a new patient-excluded shape forecast.
- usesdoi:10.18112/openneuro.ds004696.v1.0.1
The analyzed source voltages and metadata came from OpenNeuro ds004696 version 1.0.1; its saved dataset description records CC0.
- doi:10.1523/JNEUROSCI.2201-22.2023
- doi:10.1371/journal.pcbi.1008710
Cited by
6 publicationsPublic catalog entries on Sciscape.
- usesEight donor shapes gave less forecast error and 0.002 lower mean response AUC than eight cosines after five fit pulses at 86 SEEG sites in eight sessions
We use its selected voltage, pulse map, and patient-excluded shape rule. We reproduce its forecast errors and test a new response AUC endpoint.
- usesDonor response maps increased mean late SEEG forecast error by 0.24 percent in eight patient tests with five local calibration pulses
Uses its exact selected bipolar waveforms, first five pulse readout choices, event order, pinned source hashes, metadata, and retrieval code. The new test transfers an early to late voltage map across patients.
- usesA 25 ms search raised window AUC under uniform simulated ±25 ms errors at 86 sites in eight SEEG sessions
This publication supplied the exact 86 selected voltage files, pulse maps, fixed recording pairs, source metadata, and source retrieval code lineage. We fitted new templates and calculated new time-search scores. We did not reuse forecast scores.
- usesEight learned shapes with a retained residual reduced SEEG forecast error in four of eight people at 86 selected sites
The study uses its excluded-person donor-basis design. Its compact-template error motivated retaining the orthogonal residual.
- usesEight learned shapes increased mean relative SEEG prediction error with two fit pulses in five offline sessions at 20–500 ms
Its eight-shape comparison in a different cohort informed the fixed rank and the full-waveform comparison. The present test uses other shafts and two local fit pulses.
- usesFour cosine groups increased mean patient 20–500 ms forecast error by 0.31% versus one scale at 86 selected SEEG sites after five fit pulses
Provided the earlier full-waveform reliability scale and fixed cosine forecast comparison. The new rule fits four separate scales on fixed coefficient groups.