Zero-centered 8-bit waveform codes reduced frozen SEEG task-state accuracy from 70.7% to 50.6% in eight recordings
We tested voltage coding before spectral feature extraction in OpenNeuro ds004473 version 1.0.2. Each of eight people had a frozen eight-pair SEEG decoder. Native samples gave 70.69% mean balanced accuracy on later trials. A zero-centered 8-bit code with early training ranges gave 50.60%. A full-source range gave 51.03%. The planned range contrast was -0.43 percentage points, with a 95% participant bootstrap interval of -3.39 to +2.64. A post hoc test removed a fixed training median before coding and set a residual range from early samples. It gave 70.29% accuracy and changed 3.32% of native decisions. Its mean difference from native was -0.40 points, with an interval of -1.09 to +0.34. This is an exploratory within-recording task-state result. It does not prove equivalence or hardware benefit. The package preserves native bipolar windows, source retrieval records, code, models, predictions, figures, hashes, and repeat instructions.
stereoelectroencephalography · waveform quantization · signal processing · brain-computer interfaces · task-state decoding · balanced accuracy
References
4- usesFour bits per power feature gave 70.67% mean balanced accuracy in fixed SEEG tests on later trials from eight people sci:sha256:c88fa5f6642ccb1aa2a1f32f5d0ef4c199c96360c7af5e74f994a376347cd1c8
The earlier test coded power features after spectral extraction. It informed the question about a different processing stage. The present test codes voltage samples before mean removal and spectral extraction.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
We used the frozen eight-pair models, trial rules, original features, metadata, source code, and pinned EDF byte-range recipes. Native waveforms reproduced the selected parent power features and test probabilities.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
Source data: OpenNeuro ds004473, version 1.0.2, used for all eight recordings.
- doi:10.1088/1741-2552%2Facae0a
Cited by
10 publicationsPublic catalog entries on Sciscape.
- usesScore bounds with exact fallback saved 31.7% of packet bytes and kept 1,348 fixed decisions in eight offline SEEG tests
This work supplied exact selected digital bipolar windows, gains, pinned EDF ranges, hashes, metadata, and source reading code. Its fixed-offset result motivated the new mean-removed voltage packet test.
- usesJoint sample and bit reduction met a 2-point mean loss limit in eight fixed SEEG decoders under paired t assumptions
We used the compact native waveforms, reference features, fixed model states, source metadata, and separate voltage-code result.
- usesLinear filling gave 74.2 percent balanced accuracy for random labels in eight SEEG rest recordings with 20 percent sample gaps set by label
This work recovered the upstream exact bipolar waveforms and source range records. We used them through the direct parent and made a fresh audit of two source ranges per person.
- usesA power bound identified unchanged decisions under no clipping in eight fixed SEEG readers with uniform voltage codes
Supplied prior evidence about voltage offsets in raw bipolar codes. The present code centers each range from early training voltage.
- usesPair coding gave higher later balanced state accuracy than contact coding with early fixed readers and ranges at a uniform 64-bit sample budget in eight SEEG recordings
This study supplied the prior evidence on fixed-centered bipolar voltage coding. It set the pair-code comparison and showed why coding before contact subtraction needs a separate test.
- usesFit voltage offsets gave no clear four-bit site accuracy gain in offline tests of eight SEEG patients with frozen readers
The earlier task-state waveform-code study motivated the fit-median center and residual-quantile range. This work tests that center rule on a different public source and stimulation-site labels.
- usesAdding signed power changes within windows raised mean later log loss at C = 1 in eight SEEG recordings
This study uses its exact native digital bipolar windows for 2014 retained trials and its full earlier source range checks. It rebuilds the half-window spectra and adds fresh source checks for 16 trials and all 128 sub-11 trials.
- usesA causal filter gave higher fixed decoder accuracy than fourfold sample removal alone in eight SEEG recordings
This work supplies the exact retained native windows, source range hashes, trial order, frozen model inputs, and licenses. We retrieved added past voltage and tested fixed decoders after causal filtering and sample removal.
- usesA six-bit buffer code cut float32 payload size by 95.1% and raised mean balanced accuracy by 0.99 points in eight fixed offline SEEG decoders
Its fixed waveform-range results and training-centered follow-up informed the choice to remove the current buffer mean and send a current-buffer scale. The new code uses reduced-rate adaptive messages. No data or code from that study enters the new scores.
- usesA causal filter gave higher fixed decoder accuracy than fourfold sample removal alone in eight SEEG recordings
This work supplies the exact retained native windows, source range hashes, trial order, frozen model inputs, and licenses. We retrieved added past voltage and tested fixed decoders after causal filtering and sample removal.