Risk Group Statistics for the Models Built Using the Clinician Data
| Derivation group | Validation group | |||||
|---|---|---|---|---|---|---|
| Flu/Total (%) | LR | In Group (%) | Flu/Total (%) | LR | In Group (%) | |
| LASSO score | ||||||
| Low | 0/18 (0.0) | 0.0 | 10.3 | 3/9 (33.3) | 0.5 | 11.8 |
| Moderate | 19/67 (28.4) | 0.4 | 38.5 | 7/25 (28.0) | 0.4 | 32.9 |
| High | 69/89 (77.5) | 3.4 | 51.1 | 29/42 (69.0) | 2.1 | 55.3 |
| Conditional inference tree (manual) | ||||||
| Low | 0/0 (NA) | NA | 0.0 | 0/0 (NA) | NA | 0.0 |
| Moderate | 9/58 (15.5) | 0.2 | 33.3 | 8/24 (33.3) | 0.5 | 31.6 |
| High | 79/116 (68.1) | 2.1 | 66.7 | 31/52 (59.6) | 1.4 | 68.4 |
| Naive Bayes classifier | ||||||
| Low | 4/36 (11.1) | 0.1 | 20.7 | 1/9 (11.1) | 0.1 | 11.8 |
| Moderate | 3/14 (21.4) | 0.3 | 8.0 | 2/7 (28.6) | 0.4 | 9.2 |
| High | 81/124 (65.3) | 1.8 | 71.3 | 36/60 (60.0) | 1.4 | 78.9 |
Abbreviations: LASSO, Least Absolute Shrinkage and Selection Operator; LR, Stratum-specific likelihood ratio.
Notes: The models were trained using the derivation set of clinician-reported symptom data, and evaluated on both the derivation and validation sets separately. We obtained quantitative risk predictions for each individual from the models, and assigned individuals with a risk less than 10% to the low risk group, individuals with a risk between 10% and 50% to the moderate risk group, and individuals with a risk greater than 50% to the high risk group.