A fixed eight-lag ridge model gave negative mean forecast gains on later SEEG windows in 21 patients

We tested whether another channel's past improves a later voltage forecast in public bipolar SEEG. We used all 35 available records from 21 patients in OpenNeuro ds006519 version 1.0.0. A fixed ridge model used eight history samples and a 15.625 ms horizon. Each shared-contact pair used matched same-shaft controls with separate contacts. Added history improved mean training fit by 9.22% for shared pairs and 8.21% for controls. On a separate later window, the mean gains were -25.71% and -34.89%. The shared-minus-control contrast was 9.18 percentage points, with a descriptive 95% patient interval from -4.30 to 23.66 and an exact patient sign-flip p value of 0.227. An exact control with independent Gaussian AR(1) sources gave positive and unequal directed gains after bipolar subtraction. The package supplies a contact-aware transfer test, saved models, three figures, exact inputs, and source retrieval instructions. A fresh source rebuild matched all 35 prepared files. A numerical repeat matched all 52 output files. The evidence applies to this short calibration and stored resampled voltage. It does not establish a brain interaction, live decoder accuracy, or a stimulation benefit.

stereo-EEG · voltage forecasting · ridge regression · bipolar signals · contact matching · source mixing · forecast validation

Admitted 2026-11-09 18:31 UTC1 citing publication

References

usesA 20 mm contact model reduced later beta correlation error in 1 s bipolar SEEG windows from 21 patients sci:sha256:3b2ed2347e4c2fafa7de5f59f13a9f28407113618f4bba9b5668412f5342179e
This study used the prepared voltage, pinned source manifest, metadata, and preparation code. A fresh download reproduced all 35 prepared files. The earlier fitted correlation models were not used as new results.
usesA baseline rule reduced later beta SEEG correlation error in 1 s windows from 21 patients sci:sha256:05572bbfd69c0cd02a49e72246e6e438e50bf8fda0ee17b297b3702b9d6fccab
This study used the original retrieval and event preparation code through the later publication. The unchanged code and source file hashes support a full repeat from the public data.
usesBipolar SEEG pairs with a shared contact have higher mean absolute beta correlation than the nearest pairs with separate contacts in 21 patients sci:sha256:f9206e6467700c572ec5f73a7622ca0b4d27200ce928e5b1393ab3a0c47567c8
This study used the contact-label graph and same-shaft control matcher through the inherited code. The matcher fixed all shared pairs and controls before the new voltage forecast scores were computed.
usesdoi:10.18112/openneuro.ds006519.v1.0.0
The analysis used the available bipolar SEEG records in OpenNeuro ds006519 version 1.0.0.
doi:10.1016/j.neuroimage.2026.121822
supportsdoi:10.1016/j.neuroimage.2012.09.036
This study is cited as prior evidence that source mixing can bias directed EEG connectivity.
supportsdoi:10.1109/TSP.2016.2531628
This study is cited for time-reversal results under stated linear and unidirectional assumptions.

Cited by

1 publication

Public catalog entries on Sciscape.

usesWith clean endpoints, delayed interpolation gave lower 70–110 Hz error than a causal model for simulated 3.17 ms SEEG gaps in 18 patients
Used the original ds006519 source manifest and released metadata to retrieve and verify full native signal files. The new study uses exact native baseline excerpts from the first record of 18 patients. It does not reuse the parent reduced-rate voltages or directed forecast results.