Skip to main content
. 2019 Jun 4;9:8269. doi: 10.1038/s41598-019-44645-x

Table 1.

Results of the online asynchronous BCI speller.

Subject Matrix-Layout (lexicographic) QWERTZ-Layout (case-sensitive copy-spelling)
CT/min Accuracy [%] ITR [bit/min] Time [s] NC errors/min CL/min CT/min Accuracy [%] ITR [bit/min] Time [s]
S01 25.3 99.5 125.9 2.35 0 6.2 9.5 95.2 52.4 5.99
S02 34.6 99.0 170.9 1.72 0 30.7 35.5 100.0 205.0 1.69
S03 27.3 99.5 136.2 2.19 0 21.8 25.2 100.0 145.6 2.38
S04 15.9 100.0 80.0 3.76 0 7.6 9.8 94.3 53.2 5.80
S05 30.5 98.4 151.1 1.94 0.25 24.3 28.0 100.0 161.9 2.14
S06 19.7 99.5 99.5 3.02 0 11.7 13.5 97.8 76.2 4.34
S07 32.4 99.5 161.5 1.84 0.5 19.5 24.5 94.2 133.4 2.31
S08 15.3 99.5 76.2 3.91 0 4.6 5.3 100.0 30.8 11.26
S09 19.7 99.5 99.0 3.02 0 13.6 15.7 100.0 90.9 3.81
S10 25.7 99.0 126.9 2.31 0 21.3 24.5 100.0 141.8 2.45
mean 24.7 99.3 122.7 2.61 0.075 16.1 19.2 98.2 109.1 4.22

Shown are the results for both the lexicographic spelling (matrix-layout, 32 targets, 192 trials) and the case-sensitive copy-spelling (German QWERTZ-layout, 55 targets, spelling 3 times “Asynchron BCI”). For both the number of correct targets (CT) per minute, the target prediction accuracy, the information transfer rates (ITR) and the average trial duration (including an inter trial time of 0.5 s) are given. Using the matrix-layout, several non-control (NC) states were tested (4 min in total), the results are given as the average number of erroneous classifications per minute during the NC state. Additionally, for the copy-spelling, the number of correct letters (CL) per minute is given, whereas CL takes corrections and case-sensitive letters into account. Best results are in bold font.