Table 3.
Summary of classification results and quality assessment for clusters obtained from longitudinal data.
| Case | Method | Optimum Nr of clusters | Majority baseline | Accuracy (f1_macro) RF over-sampling |
95% CI for accuracy | Clinical sensibleness |
|---|---|---|---|---|---|---|
| Physical Health | kml3d | 8 | 16.48 | 51.89 (50.03) | [50.84–52.94] | +++ |
| Psychological Health | kml3d | 9 | 26.15 | 83.12 (82.63) | [82.36–83.87] | ++ |
| General Health | kml3d | 10 | 17.07 | 68.26 (68.02) | [67.66–68.85] | +++ |
| Physical Health (pre-injury) | kml3d | 8 | 18.08 | 61.56 (60.74) | [60.32–62.81] | ++ |
| Physical Health | HDclassif | 7 | 24.49 | 69.52 (68.61) | [68.00–71.05] | +++ |
| Psychological Health | HDclassif | 10 | 19.12 | 70.24 (69.75) | [68.78–71.70] | + |
| General Health | HDclassif | 6 | 30.03 | 73.96 (72.59) | [72.89–75.03] | +++ |
| Physical Health (pre-injury) | HDclassif | 6 | 26.07 | 69.12 (68.64) | [67.81–70.44] | +++ |
| Physical Health | Deepgmm | 6 | 45.32 | 91.30 (90.85) | [90.50–92.11] | +++ |
| Psychological Health | Deepgmm | 6 | 84.70 | 99.96 (98.67) | [99.94–99.98] | + |
| General Health | Deepgmm | 6 | 62.13 | 98.20 (97.87) | [97.87–98.54] | ++ |
| Physical Health (pre-injury) | Deepgmm | 6 | 61.02 | 94.78 (93.55) | [94.37–95.18] | + |
Best models based on classification metrics and clinical sensibleness are in bold.