← All interfaces

Synchron Stentrode, switch (Oxley et al., 2021)

Measured by Oxley, Yoo, Opie et al. (first-in-human study) · J. NeuroInterventional Surgery 13(2) (2021)

Endovascular Switch partially-invasive 2021

Inputs

The measured or assumed values behind the calculations, each with its source.

N = 28
On-screen keyboard targets reached with the switch (letters + space + edits, approx.)
rate = 13.81 char/min
Participant 1 correct characters/min, predictive text disabled (first-in-human typing task; Participant 2 reached 20.10 cpm)
H = 1.0 bits/char
English-text entropy (Shannon); the task produced English text, so the same ~1 bit/char standard used for QWERTY, eye-typing and the BrainGate2 typing entry applies.
P = 0.9263
Participant 1 average click-selection accuracy, 92.63% (first-in-human typing task)

Strictest ITR

Each scoring method is an upper bound on the channel, so the headline is the strictest (smallest) one for this entry. Use the score selector on the home page to view any single method across entries.

Strictest Shannon (text) Recomputed
Character-entropy throughput (realized text entry)
Net of English redundancy
0.23 bits/s
  1. Characters per minute

    13.81 correct char/min (Participant 1, predictive text disabled; P2 reached 20.10 cpm)

    The eye-tracker supplies the pointing and the endovascular electrode supplies the click; this is the rate of English text actually produced. The same char-entropy method as QWERTY, eye-typing and the BrainGate2 typing entry.

  2. Bits per character

    H(English) ≈ 1.0 bit/char (Shannon)
  3. Information transfer rate

    13.81 char/min × 1.0 bit/char ÷ 60 s/min = 0.23 bits/s

What counts as a bit depends on the action space. The number of distinguishable actions and how likely each one is are design choices of the task, not the sensing hardware. The same modality can present a fixed set of targets, a set pruned per step by a grammar or language model, or a continuous control space. Each of these changes how many actions are live and how the probability mass is spread, and therefore the information per selection. Read the action space below before comparing headline numbers across entries.

Action space

What the user can produce at each step, and how those options are distributed.

Structure
Context-dependent (the live set changes per step)
Size
28 distinguishable actions
Prior
Context-conditioned: likelihoods depend on prior actions
Notes
Credit-assignment caveat: the endovascular electrode supplies a switch signal; spatial pointing is supplied by an eye-tracker, and a scanning/keyboard interface gates the targets. This per-character figure therefore measures the complete assistive stack rather than the endovascular signal alone. Because the task produced English text under a context-conditioned prior, the reference number uses the same character-entropy method as the other text-entry entries (~1 bit/char); the Wolpaw-over-28-targets figure, which assumes a uniform 1-of-28 choice, is kept only as a secondary selection metric.

Comparability The strictest bound here is the Shannon entropy of the output text, under one predictor held constant across the whole atlas (≈1 bit per character). That shared predictor makes it directly comparable to every other text entry (keyboards, spellers, silent speech and speech BCIs) regardless of prior or vocabulary size. For most text interfaces it comes out tighter than the raw-selection bounds, but not always. Where a small vocabulary makes Wolpaw tighter, that wins instead. Any Fitts, Wolpaw or log₂(N) figure shown below is another bound on the same channel. Switch the home-page score selector to compare one across entries.

Other score types

Bounds the atlas keeps out of the default strictest headline: as-reported figures, alternate task conditions, or raw-channel ceilings that shouldn't win the headline by default. Each still carries a score type, so the home-page selector ranks this entry on it when you choose that type. Read its derivation before comparing across entries.

Wolpaw Recomputed
Wolpaw bitrate over N = 28 keyboard targets
Uniform-prior selection metric, shown for comparison
0.938 bits/s
  1. Bits per selection (Wolpaw formula)

    B = log2(N) + P*log2(P) + (1-P)*log2((1-P)/(N-1))
      = log2(28) + 0.9263*log2(0.9263) + 0.0737*log2(0.0737/27)
      = 4.077 bits / selection

    Term 1 is the information if every choice were correct; terms 2-3 subtract the bits lost to the error rate, assumed spread evenly over the other N-1 targets.

  2. Selections per second

    T = 4.34468 s/selection  ->  1 / 4.34468 = 0.23 selections/s
  3. Information transfer rate

    ITR = B * selections/s = 4.077 * 0.23 = 0.938 bits/s
Nuyujukian Recomputed
Nuyujukian achieved bitrate over N = 28 keyboard targets
Achieved-bitrate view of the selection channel, shown for comparison
0.93 bits/s
  1. Achieved-bitrate credit per net-correct selection

    N = 28 → log2(N − 1) = log2(27) = 4.75 bits per net-correct selection (field-standard achieved bitrate, e.g. Webgrid; Nuyujukian 2015, which introduced the metric, used log2(N)).
  2. Net-correct selection rate

    net-correct = 2P − 1 = 2(0.9263) − 1 = 0.853 of selections. At 4.34 s/selection → 0.853 / 4.34 = 0.196 correct/s.

    A wrong click commits the wrong key rather than timing out, so incorrect = 1 − P. Same N (28), Participant 1 click accuracy (92.63%) and selection interval (4.34 s) as the entry's Wolpaw calc; netting each wrong click against a correct one (2P − 1) lands just under the ~0.94 bits/s Wolpaw figure. Like it, this measures the whole assistive stack (eye-tracker pointing + endovascular click), not the endovascular signal alone.

  3. Achieved bitrate

    4.75 bits × 0.196 correct/s = 0.93 bits/s.

Source

Authors
Oxley, Yoo, Opie et al. (first-in-human study)
Publication
J. NeuroInterventional Surgery 13(2), 2021
Paper
10.1136/neurintsurg-2020-016862
Reference
Mitchell et al. 2023 (JAMA Neurology): SWITCH trial 4-patient safety outcomes