AUC | Samples per subject | Controls | Crohn’s disease | Sampling |
0.80 | 1* | 12 | 19 | Daily samples |
0.77 | 1 | 12 | 19 | |
0.82 | 2 | 12 | 19 | |
0.85 | 3 | 12 | 18 | |
0.86 | 4 | 12 | 18 | |
0.87 | 5 | 12 | 18 | |
0.87 | 6 | 12 | 18 | |
0.88 | 7 | 12 | 18 | |
0.88 | 8 | 12 | 18 | |
0.87 | 9 | 12 | 18 | |
0.87 | 10 | 12 | 17 | |
0.86 | 11 | 12 | 16 | |
0.80 | 1* | 9 | 19 | Monthly samples3 |
0.80 | 1 | 9 | 19 | |
0.83 | 2 | 9 | 15 | |
0.86 | 3 | 8 | 14 | |
0.92 | 4 | 8 | 12 |
The AUC summarises the performance; closer to 1 is better, of the model trained on the different sample sizes as described by the other columns.*Represents the performance of a classifier that relies on non-longitudinal relative abundances only.AUC, area under the curve.