Table 2. Group- and individual-level results of classification [predicting clock time–matched >24 hours awake (0 to 16 hours to 24 to 38 hours awake)] and regression (predicting TSW) random forest models for within-participant and between-participant analyses.
Showing models trained on experiment 1 (tested experiment 2 group and participants) using the five final biomarker candidates. For classification models, model accuracy (lower–upper 95% CI), area under the receiver operating curve (AUC %, lower–upper 95% CI), negative and positive prediction accuracy (NPV and PPV%), specificity (SP%), and sensitivity (SN%) are displayed (tree depth was 35 and 53, for within-participant and between-participant classification training models, respectively). Note that if PPV is 0, SN cannot be calculated. For regression models, variance explained (as R2) and RMSE are displayed (tree depth was 159 and 171, for within-participant and between-participant regression training models, respectively).
| Comparison | Model | Participant | Accuracy (%) (lower–upper) | AUC (%) (lower–upper) | SP (%) | SN (%) | NPV (%) | PPV (%) | R2 (%) | RMSE |
|---|---|---|---|---|---|---|---|---|---|---|
| Within participant | Training | Group | 97.8 (94.5–99.4) | 99.7 (99.3–100) | 98.9 | 96.7 | 96.8 | 98.9 | 98.0 | 1.52 |
| Testing | Group | 94.7 (90.1–97.5) | 99.2 (98.3–100) | 96.3 | 93.1 | 92.9 | 96.4 | 86.7 | 4.06 | |
| M | 100 (79.4–100) | 100 (100–100) | 100 | 100 | 100 | 100 | 90.1 | 3.45 | ||
| N | 93.8 (69.8–99.8) | 100 (100–100) | 88.9 | 100 | 100 | 87.5 | 89.4 | 3.57 | ||
| O | 93.8 (69.8–99.8) | 95.3 (85.1–100) | 88.9 | 100 | 100 | 87.5 | 86.1 | 4.09 | ||
| P | 81.2 (54.4–96) | 95.3 (86.2–100) | 100 | 72.7 | 62.5 | 100 | 57.7 | 7.13 | ||
| Q | 93.8 (69.8–99.8) | 100 (100–100) | 88.9 | 100 | 100 | 87.5 | 93.4 | 2.90 | ||
| R | 100 (79.4–100) | 100 (100–100) | 100 | 100 | 100 | 100 | 93.6 | 2.77 | ||
| S | 100 (79.4–100) | 100 (100–100) | 100 | 100 | 100 | 100 | 84.6 | 4.30 | ||
| T | 100 (79.4–100) | 100 (100–100) | 100 | 100 | 100 | 100 | 85.3 | 4.20 | ||
| U | 92.9 (66.1–99.8) | 100 (100–100) | 100 | 87.5 | 85.7 | 100 | 87.4 | 4.49 | ||
| V | 92.9 (66.1–99.8) | 100 (100–100) | 100 | 87.5 | 85.7 | 100 | 90.5 | 3.33 | ||
| W | 92.3 (64–99.8) | 100 (100–100) | 100 | 85.7 | 85.7 | 100 | 93.5 | 2.84 | ||
| Between participant | Training | Group | 92.9 (88.1–96.1) | 97.7 (95.8–99.6) | 94.4 | 91.3 | 91.4 | 94.4 | 94.0 | 2.66 |
| Testing | Group | 79.3 (72.4–85.1) | 89.1 (84.4–93.8) | 76.0 | 83.6 | 85.9 | 72.6 | 48.2 | 8.00 | |
| M | 75 (47.6–92.7) | 100 (100–100) | 66.7 | 100 | 100 | 50.0 | 51.9 | 7.60 | ||
| N | 100 (79.4–100) | 100 (100–100) | 100 | 100 | 100 | 100 | 74.6 | 5.52 | ||
| O | 68.8 (41.3–89) | 90.6 (75.6–100) | 100 | 61.5 | 37.5 | 100 | 28.6 | 9.25 | ||
| P | 50 (24.7–75.3) | 96.9 (89.6–100) | 50.0 | 100 | 0.0 | −28.4 | 12.41 | |||
| Q | 87.5 (61.7–98.4) | 100 (100–100) | 100 | 80.0 | 75.0 | 100 | 65.0 | 6.65 | ||
| R | 100 (79.4–100) | 100 (100–100) | 100 | 100 | 100 | 100 | 88.9 | 3.64 | ||
| S | 50 (24.7–75.3) | 94.5 (84–100) | 50.0 | 100 | 0.0 | −9.3 | 11.45 | |||
| T | 81.2 (54.4–96) | 100 (100–100) | 72.7 | 100 | 100 | 62.5 | 56.5 | 7.23 | ||
| U | 100 (76.8–100) | 100 (100–100) | 100 | 100 | 100 | 100 | 66.0 | 7.37 | ||
| V | 85.7 (57.2–98.2) | 90.8 (73.7–100) | 100 | 77.8 | 71.4 | 100 | 69.1 | 6.00 | ||
| W | 84.6 (54.6–98.1) | 95.2 (84.2–100) | 100 | 75.0 | 71.4 | 100 | 65.3 | 6.59 |