• Table 1.

    Survey of BCI studies that measure typing rates (correct characters per minute; ccpm), bitrates, or information transfer rates for people with motor impairment. Number ranges represent performance measurements across all participants for a given study. Communication rates could be further increased by external algorithms such as word prediction or completion. As there are many such algorithms, the current work excluded word prediction or completion to focus on measuring the performance of the underlying system. The most appropriate points of comparison, when available, are bitrates, which are independent of word prediction or completion algorithms. Similarly, information transfer rates are also a meaningful point of comparison, though they are less reflective of practical communication rates than bitrate (which takes into account the need to correct errors; detailed in Nuyujukian et al. (2015); Townsend et al. (2010)). For the current work, and for Jarosiewicz et al. 2015, we also break down performance by individual participant to facilitate direct comparisons (denoted by italics). As shown, performance in the current study outperforms all previous BCIs tested with people with motor impairment. *These numbers represent performance when measured using a denser grid (9 × 9; Figure 3—figure supplement 2 and Video 10). **For this study, reported typing rates included word prediction / completion algorithms. ***Number range represents the range of performance reported for the single study participant. ****Other reported numbers included word prediction / completion algorithms. †Acronyms used: ReFIT-KF: Recalibrated Feedback Intention-trained Kalman Filter. HMM: Hidden Markov Model. CLC: Closed-loop Calibration. LDA: Linear Discriminant Analysis. RTI: Retrospective Target Inference. DS: Dynamic Stopping.

    DOI: http://dx.doi.org/10.7554/eLife.18554.021

    StudyParticipant(s)Recording modalityControl modalityEtiology of
    motor impairment
    Average typing rate (ccpm)Average bitrate (bps)Average ITR (bps)
    This studyaverage
    (N = 3)
    intracorticalReFIT-KF+HMMALS (2), SCI (1)28.12.42.4
    ‘‘T6ALS31.62.22.2
    ‘‘T5SCI39.23.73.7
    ‘‘‘‘‘‘-4.2*4.2*
    ‘‘T7(No HMM)ALS13.51.41.4
    Bacher et al., 2015S3intracorticalCLC+LDAbrainstem stroke9.4--
    Jarosiewicz et al., 2015average (N = 4)intracorticalRTI+LDAALS (2),
    brainstem stroke (2)
    n/a**0.59-
    ‘‘T6ALS‘‘0.93-
    ‘‘T7ALS‘‘0.64-
    ‘‘S3brainstem stroke‘‘0.58-
    ‘‘T2brainstem stroke‘‘0.19-
    Nijboer et al., 2008N = 4EEGP300ALS1.5–4.1-0.08–0.32
    Townsend et al., 2010N = 3EEGP300ALS-0.05–0.22-
    Münßinger et al., 2010N = 3EEGP300ALS--0.02–0.12
    Mugler, et al. 2010N = 3EEGP300ALS--0.07–0.08
    Pires et al., 2011N = 4EEGP300ALS (2), cerebral palsy (2)--0.24–0.32
    Pires et al., 2012N = 14EEGP300ALS (7), cerebral palsy (5),
    Duchenne muscular
    dystrophy (1), spinal cord injury (1)
    --0.05–0.43
    Sellers et al., 2014N = 1EEGP300brainstem stroke0.31–0.93***--
    McCane et al., 2015N = 14EEGP300ALS--0.19
    Mainsah et al., 2015N = 10EEGP300-DSALS--0.01–0.60
    Vansteensel et al., 2016N = 1subdural ECoGLinear ClassifierALS1.15****-0.21
  • Table 2.

    Participants’ prior BCI experience and training for studies considered in Table 1. The experience column details the number of participants in the respective study that had prior experience with BCIs at the time of the study and, if reported, the duration of that prior experience and/or training.

    DOI: http://dx.doi.org/10.7554/eLife.18554.024

    StudyParticipant(s)BCI experience/training
    This studyaverage
    (N = 3)
    1 year
    ‘‘T61.5 years
    ‘‘T59 prior sessions (1 month)
    ‘‘T71.5 years
    Bacher et al., 2015S34.3 years
    Jarosiewicz et al., 2015average
    (N=4)
    2 years
    ‘‘T610 months to 2.3 years
    ‘‘T75.5 months to 1.2 years
    ‘‘S35.2 years
    ‘‘T24.6 months
    Nijboer et al., 2008N = 4At least 4–10 months
    Townsend et al., 2010N = 3All had prior P300 BCIs at home, two had at least 2.5 years with BCIs
    Münßinger et al., 2010N = 3Two of three had prior experience, training not reported
    Mugler, et al. 2010N = 3Average experience of 3.33 years
    Pires et al., 2011N = 4No prior experience, training not reported
    Pires et al., 2012N = 14Not reported
    Sellers et al., 2014N = 1Prior experience not reported, thirteen months of continuous evaluation
    McCane et al., 2015N = 14Not reported
    Mainsah et al., 2015N = 10Prior experience not reported, two weeks to two months of evaluation
    Vansteensel et al., 2016N = 17 to 9 months
  • Table 3.

    Summary of the decoding and calibration approaches used with each participant.

    DOI: http://dx.doi.org/10.7554/eLife.18554.032

    T6

    T7

    T5

    Continuous decoding algorithm

    ReFIT Kalman Filter (threshold crossings and HF-LFP)

    ReFIT Kalman Filter

    (threshold crossings)

    ReFIT Kalman Filter

    (threshold crossings)

    Discrete decoding algorithm

    Hidden Markov Model (HF-LFP)

    n/a

    Hidden Markov Model (threshold crossings)

    Dwell time

    1 s (reset on target exit)

    1.5 s (cumulative)

    1 s (reset on target exit)

    Bias estimation

    no

    yes

    yes

    Cursor recentering

    no

    yes

    no

    Recalibration blocks

    Recalibrated continuous and discrete decoders

    Only updated bias estimates

    Only updated bias estimates

    Error attenuation in recalibration block

    yes

    yes

    no

  • Video 1. Example of participant T6’s free-paced, free choice typing using the OPTI-II keyboard.

    T6 was prompted with questions and asked to formulate an answer de novo. Once presented with a question, she was able to think about her answer, move the cursor and click on the play button to enable the keyboard (bottom right corner), and then type her response. In this example, the participant typed 255 characters in ~9 min, at just over 27 correct characters per minute. One of two audible ‘beeps’ followed a target selection, corresponding to the two possible selection methods: T6 could select targets using either the Hidden Markov Model-based ‘click’ selection (high-pitched noises) or by ‘dwelling’ in the target region for 1 s (low-pitched noises). The plot at the bottom of the video tracks the typing performance (correct characters per minute) with respect to time in the block. Performance was smoothed using a 30 s symmetric Hamming window. The scrolling yellow bar indicates the current time of that frame. During the free typing task, T6 was asked to suppress her hand movements as best as possible. (During the quantitative performance evaluations, T6 was free to make movements as she wished.) This video is from participant T6, Day 621, Block 17. Additional ‘free typing’ examples for T6 are detailed in Figure 1—figure supplements 1 and 2.

    DOI: http://dx.doi.org/10.7554/eLife.18554.007

  • Video 2. Example of participant T6’s ‘copy typing’ using the OPTI-II keyboard.

    In the copy typing task, participants were presented with a phrase and asked to type as many characters as possible within a two-minute block. T6 preferred that the cursor remain under her control throughout the task – i.e., no re-centering of the cursor occurred after a selection. This video is from participant T6, Day 588, Blockset 2. Performance in this block was 40.4 ccpm.

    DOI: http://dx.doi.org/10.7554/eLife.18554.012

  • Video 3. Example of participant T6’s ‘copy typing’ using the QWERTY keyboard.

    Same as Video 2, but using the QWERTY keyboard layout. This video is from participant T6, Day 588, Blockset 4. Performance in this block was 30.6 ccpm.

    DOI: http://dx.doi.org/10.7554/eLife.18554.013

  • Video 4. Example of participant T5’s ‘copy typing’ using the OPTI-II keyboard.

    Same as Video 2, but for participant T5. This video is from participant T5, Day 68, Blockset 4. Performance in this block was 40.5 ccpm.

    DOI: http://dx.doi.org/10.7554/eLife.18554.014

  • Video 5. Example of participant T5’s ‘copy typing’ using the QWERTY keyboard.

    Same as Video 4, but using the QWERTY keyboard layout. This video is from participant T5, Day 68, Blockset 2. Performance in this block was 38.6 ccpm.

    DOI: http://dx.doi.org/10.7554/eLife.18554.015

  • Video 6. Example of participant T7’s ‘copy typing’ using the OPTI-II keyboard.

    Same as Video 2, but for participant T7. T7 selected letters by dwelling on targets only. In addition, T7 preferred that the cursor re-center after every selection (i.e., following a correct or an incorrect selection). These across-participant differences are detailed in Materials and methods: Quantitative performance evaluations (under ‘Target selection and cursor re-centering’). This video is from participant T7, Day 539, Blockset 3. Performance in this block was 10.6 ccpm.

    DOI: http://dx.doi.org/10.7554/eLife.18554.016

  • Video 7. Example of participant T7’s ‘copy typing’ using the ABCDEF keyboard.

    Same as Video 6, but using the ABCDEF keyboard layout. This video is from participant T7, Day 539, Blockset 1. Performance in this block was 16.5 ccpm.

    DOI: http://dx.doi.org/10.7554/eLife.18554.017

  • Video 8. Example of participant T6’s performance in the grid task.

    This video is from participant T6, Day 588, Blockset 3. Performance in this block was 2.65 bps.

    DOI: http://dx.doi.org/10.7554/eLife.18554.022

  • Video 9. Example of participant T5’s performance in the grid task.

    This video is from participant T5, Day 56, Blockset 4 (Block 28). Performance in this block was 4.01 bps.

    DOI: http://dx.doi.org/10.7554/eLife.18554.023

  • Video 10. Example of participant T5’s performance in the dense grid task (9 × 9).

    This video is from participant T5, Day 56, Blockset 4 (Block 30). Performance in this block was 4.36 bps.

    DOI: http://dx.doi.org/10.7554/eLife.18554.025

  • Video 11. Example of participant T7’s performance in the grid task.

    This video is from participant T7, Day 539, Blockset 2. Performance in this block was 1.57 bps.

    DOI: http://dx.doi.org/10.7554/eLife.18554.026