Table 5.
Comparison of peak rotational velocity (PRV) across six data cleaning methods
| Peak rotational velocity (PRV, rad/s) | |||||
|---|---|---|---|---|---|
| Method | Median | 25th percentile | 25th percentile | Compared method | Corrected p-value |
|
1 (uncleaned) |
8.5 | 6.5 | 10.1 | 2 | 1.000 |
| 3 | 0.003 * | ||||
| 4 | 0.268 | ||||
| 5 | 1.000 | ||||
| 6 | 1.000 | ||||
|
2 (time-windowed HAEs only) |
8.5 | 6.2 | 9.8 | 3 | 0.063 |
| 4 | 1.000 | ||||
| 5 | 1.000 | ||||
| 6 | 1.000 | ||||
|
3 (algorithm-classified HAEs without time windowing) |
6.5 | 4.8 | 9.4 | 4 | 1.000 |
| 5 | 0.449 | ||||
| 6 | 0.672 | ||||
|
4 (algorithm-classified HAEs with time windowing) |
7.2 | 5.3 | 10.0 | 5 | 1.000 |
| 6 | 1.000 | ||||
|
5 (video verification) |
7.2 | 6.0 | 10.4 | 6 | 1.000 |
|
6 (HAEs confirmed by both proprietary classification algorithm and video verification) |
7.4 | 5.6 | 8.8 | ||
Median peak rotational velocity (PRV) values are presented for each data cleaning method. Differences in PRV across methods were assessed using a related-samples Friedman’s Two-Way ANOVA with significance values adjusted by a Bonferroni correction. *indicates a significant difference was observed