Skip to main content
. 2020 Sep 30;14:521282. doi: 10.3389/fnhum.2020.521282

TABLE 2.

Classification performance of the different algorithms.

Algorithm Correctly classified (%) Incorrectly classified (%) Algorithm Correctly classified (%) Incorrectly classified (%)
Ranking high-to-low spectral power Ranking high-to-low spectral power
Window, moving Current, moving
Total beta 83.3 16.7 Total beta 16.7 83.3
Low beta 58.3 41.7 Low beta 16.7 83.3
High beta 66.7 33.3 High beta 25 75
Peak beta 58.3 41.7 Peak beta 16.7 83.3
Window, resting Current, resting
Total beta 83.3 16.7 Total beta 41.7 58.3
Low beta 66.7 33.3 Low beta 50 50
High beta 41.7 58.3 High beta 25 75
Peak beta 58.3 41.7 Peak beta 16.7 83.3
Window, moving-to-resting Current, moving-to-resting
Total beta 41.7 58.3 Total beta 25 75
Low beta 25 75 Low beta 41.7 58.3
High beta 33.3 66.7 High beta 16.7 83.3
Peak beta 33.3 66.7 Peak beta 16.7 83.3
Ranking low-to-high spectral power
Current, moving
Total beta 66.7 33.3
Low beta 66.7 33.3
High beta 58.3 41.7
Peak beta 66.7 33.3
Current, resting
Total beta 41.7 58.3
Low beta 58.3 41.7
High beta 33.3 66.7
Peak beta 41.7 58.3
Current, moving-to-resting
Total beta 83.3 16.7
Low beta 66.7 33.3
High beta 41.7 58.3
Peak beta 66.7 33.3

Best performing classifiers are set in bold and are also shown in Figure 5. Left part of the table using the therapeutic window for the actual ranking (larger window—higher ranking); right part of the table using the therapeutic current threshold for the actual ranking (lower threshold—higher ranking).