Table 1.
Categorization accuracy and statistical analysis for the scenarios Rest 1, Math, and Exercise across all ten subjects in Part 1.
| Metric | M1 | M2 | M3 | |||
|---|---|---|---|---|---|---|
| CA (%) | pR−M | pR−E | pM−E | CA (%) | CA (%) | |
| LFp | 58.4 | 0.52 | 0.24 | 0.12 | 46.8 | 80.4 |
| HFp | 52.1 | 0.47 | <0.01 | 0.06 | 59.5 | 88.6 |
| LFp/HFp | 48.8 | 0.16 | 0.16 | 1.00 | 52.1 | 79.9 |
| 2Dp | 76.7 | nA | nA | nA | 75.7 | 98.3 |
| LFn | 44.7 | 0.31 | 0.24 | 0.68 | 55.5 | 75.7 |
| HFn | 50.7 | 0.09 | 0.09 | 0.91 | 57.5 | 80.3 |
| LFn/HFn | 48.4 | 0.16 | 0.16 | 1.00 | 52.1 | 79.9 |
| 2Dn | 61.7 | nA | nA | nA | 71.2 | 92.0 |
| LFiA | 40.9 | 0.16 | 0.08 | 0.04 | 57.8 | 78.3 |
| HFiA | 57.7 | 0.14 | <0.01 | 0.12 | 65.3 | 94.0 |
| LFiA/HFiA | 54.7 | 0.03 | 0.02 | 0.31 | 56.7 | 83.2 |
| 2DiA | 75.1 | nA | nA | nA | 85.9 | 99.6 |
Categorization based on the raw, absolute, stress parameters is given in column M1, while panel M2 shows the values after standardizing (for each subject separately) by subtracting the median values of the “baseline” segment Rest 1 from the whole recording (this improved cross-subject comparisons), and column M3 shows the values produced with the categorization performed for all subjects individually. The statistical differences were calculated for one-dimensional parameters and denoted as follows: pR−M: Rest 1 and Math, pR−E: Rest 1 and Exercise, pM−E: Math and Exercise, while the two-dimensional parameters are denoted with 2Dj, j∈{p, n, iA}. The values in bold are the categorisation accuracies for the proposed 2D analysis.