Adding signed power changes within windows raised mean later log loss at C = 1 in eight SEEG recordings
We tested power changes within 500-sample windows from eight public SEEG recordings. Models used early trials to predict rest and the state before a response in later trials. The added signed changes raised mean log loss from 0.752 to 0.854 nats per window at fixed C = 1. Loss rose in seven of eight people. Mean balanced accuracy changed from 71.70% to 71.88%. A planned stronger penalty, C = 0.1, gave mean losses of 0.561 and 0.560. This result concerns fixed logistic models, event-aligned task states, and reused later trials. It does not establish that temporal information is absent. The package contains exact samples, code, source records, models, predictions, figures, and checks.
stereoelectroencephalography · brain-computer interface · signal processing · power features · logistic regression · log loss · temporal features · probability forecasting
References
4- usesZero-centered 8-bit waveform codes reduced frozen SEEG task-state accuracy from 70.7% to 50.6% in eight recordings sci:sha256:f11a99582db0d0fff28986ef433b9cde1661cb1668cb5772f0a78262582c2b8c
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.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
This study keeps the eight selected bipolar pairs, early and later trial split, reference power values, and original model record. The later trials are reused. The new test adds signed half-power differences to fixed logistic models.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis uses the eight recordings and exact waveform windows from OpenNeuro ds004473 version 1.0.2; the archived dataset metadata identifies this version and its CC0 license.
- doi:10.1088/1741-2552%2Facae0a
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesSigned power changes reduced mean response phase log loss with early setting selection in eight offline SEEG tests
Used its F/M/A/D feature definitions and fixed-C result to define the ordered-feature comparison. This test changes the target to pre/post response phase and selects C from early trials.