Two-second interval thresholds reduced background output and detection before keypresses in eight later SEEG records
We tested thresholds for any output during two-second background intervals in eight public SEEG records. The fixed decoder and continuous powers came from earlier work on OpenNeuro ds004473 version 1.0.2. Early background intervals set a nominal five percent threshold. Later intervals tested it. A threshold from all window scores gave 31.4 percent mean interval output. A threshold from interval maxima gave 3.7 percent. Mean detection in the half second before recorded keypresses fell from 43.4 to 10.2 percent. An equal-count last-window control gave 24.4 percent interval output. The main test had 727 later background intervals and 425 keypresses. Six patient rates were below five percent; two were 5.04 and 8.43 percent. A known rank rule gives a bound for one future interval only under exchangeability. The time-ordered records do not prove that assumption. The package preserves inputs, weights, timing, code, figures, source versions, hashes, and reproduction commands. Fresh public raw checks covered 5,520 windows. These offline results show the risk unit and the detection cost. They do not establish user intent or test brain stimulation.
brain-computer interface · stereoelectroencephalography · threshold calibration · conformal prediction · background output · keypress detection
References
6- usesThree consecutive scores gave no clear pre-response detection gain in eight SEEG streams at an early target of one command per eligible minute sci:sha256:0b8e19ef6848ab50266578c54c030f4887da0418334c3bdb63195df9f582dc33
Defines prior work on threshold fitting for commands per eligible minute and consecutive scores. The new comparison tests any candidate output per monitored interval and its detection cost.
- usesFixed SEEG power models gave no clear Brier score gain beyond cue and response history inputs for 100 ms keypress forecasts in eight later task segments sci:sha256:bbeeb9b787afdd035adf78c2e020da8123883495d8f8654c4b6aa29932ff6ef2
Supplies the copied continuous powers, fixed ridge weights, event files, pinned public retrieval records, and imported raw check script. This package checks the hashes of all 76 copied files.
- usesAt a zero threshold, eight-pair SEEG decoders gave a mean of 36.7 outside-guard triggers per eligible minute in eight later task scans sci:sha256:22495a59283fcfea65785cbf2a7a410c3a81dc75e721320a08a5ed114e14bd43
Created the continuous eight-pair SEEG streams and outside-response guard convention. The new result uses candidate-output probability per fixed interval rather than refractory trigger rate per minute.
- usesRest calibration gave a 4.2 percent mean false positive rate in eight later-trial SEEG tests sci:sha256:7a28b8fc519f8f608fe5546ba40889db9bb37b5c93f68ed3821b4ba5bf1c5e75
Defines prior work on a five percent rank threshold for single rest windows. The new test applies the known rank rule to continuous interval maxima with a different fixed feature set.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis used the eight clinical task records in OpenNeuro ds004473 version 1.0.2. The source package preserves pinned EDF versions, byte ranges, and hashes.
- doi:10.1088/1741-2552%2Facae0a
Cited by
4 publicationsPublic catalog entries on Sciscape.
- supportsFixed SEEG power models gave no clear gain beyond task events for continuous 100 ms keypress forecasts in eight people
This related work showed that the time unit changes continuous output rates. It used two-second background intervals and an inherited task-state score. The present test uses a 100 ms forecast and a causal task reference.
- usesPast averaging of three scores gave 9.0% mean detection before keypresses at a 5% interval target in eight SEEG records
This study reuses its fixed SEEG powers, reader models, event records, source metadata, raw range recipes, and independent Fourier audit code. It compares past score rules after each rule receives an early interval threshold.
- usesA fixed monitor warned in eight SEEG replays with output above 5%; its warning bound requires a conditional rate limit
This study uses the continuous SEEG powers, fixed score models, event arrays, source interval labels, and pinned raw retrieval records. It tests a new sequential monitor in the same eight people. It rebuilds the primary source labels before the new replay.
- usesRates of any reader output over two seconds were below expected rates for uniform random orders with fixed positive counts in eight SEEG records
This study used the compact SEEG powers, fixed reader weights, events, range recipes, and raw audit code. It reproduced the published baseline rate.