Table 4.
Model # | Stability (%) | Average accuracy (%) | COV (%) | |||
---|---|---|---|---|---|---|
Cohort | YS | OS | YS | OS | YS | OS |
1 | < 1* | < 1* | − 12* | − 12*** | 11* | 17* |
2 | − 41 | − 76 | 429 | − 27 | 89 | 86 |
3 | 16 | − 43 | 280 | − 16 | 152 | 85 |
4 | < 1* | < 1* | 16** | 8* | 31*** | 37*** |
5 | 212 | 258 | 150 | 90 | 96 | 85 |
6 | − 72 | − 31 | − 75 | − 29 | 108 | 105 |
7 | < 1* | < 1* | − 29*** | − 11** | 32 | 39 |
8 | 53 | 2435 | 163 | 2546 | 132 | 156 |
9 | − 22 | − 65 | − 30 | − 68 | 15** | 64 |
10 | − 32 | − 10 | − 31 | − 81 | 53 | 27** |
Stability: maximum percent difference between the oxidation parameter derived by fitting the noiseless curve and the ground truth. Average accuracy: maximum percent difference between the average oxidation parameter derived by fitting 100 noisy curves and the ground truth. COV coefficient of variation (standard deviation/mean) for 100 noisy curve fits. For each metric and each cohort, the 3 best (lowest) values are indicated (*best result, **second best, ***third best). YS younger subjects, OS older subjects