TABLE 7.
Classification performance of a predictive multivariable logistic regression model at different thresholds of predicted probability
| Predicted probability (%) | PPV (%) | NPV (%) | Sensitivity (%) | Specificity (%) |
|---|---|---|---|---|
| 10 | 38.90 (37.24, 40.64) | 95.51 (92.91, 97.72) | 95.00 (92.08, 97.50) | 41.12 (37.13, 45.07) |
| 20 | 47.50 (44.62, 50.63) | 89.93 (87.42, 92.34) | 81.67 (77.67, 86.25) | 64.31 (60.69, 67.93) |
| 30 | 55.21 (51.23, 59.41) | 86.90 (84.63, 89.28) | 70.42 (64.58, 70.42) | 77.47 (74.18, 80.60) |
| 40 | 62.29 (57.20, 67.53) | 84.57 (82.46, 86.69) | 60.42 (54.17, 66.68) | 85.53 (82.57, 88.32) |
| 50 | 71.59 (65.88, 77.38) | 83.51 (81.57, 85.45) | 54.17 (47.50, 60.42) | 91.45 (89.14, 93.59) |
| 60 | 78.29 (71.65, 84.56) | 80.89 (79.18, 82.65) | 42.92 (36.37, 49.58) | 95.25 (93.59, 96.88) |
| 70 | 86.25 (79.00, 92.96) | 78.20 (76.82, 79.71) | 30.83 (25.00, 37.08) | 98.03 (96.88, 99.01) |
| 80 | 86.11 (75.00, 95.12) | 74.69 (73.71, 75.72) | 15.00 (10.42, 19.58) | 99.01 (98.19, 99.67) |
| 90 | 100 (100, 100) | 72.73 (72.21, 73.34) | 5.00 (2.50, 7.92) | 100 (100, 100) |
Notes: Incremental increases in the predicted probability were evaluated as the threshold used to classify dogs as being in stage B2. The utility of each threshold was assessed using training set data. Confidence intervals were calculated using 2000 stratified bootstrap replicates (R package “pROC” v1.16.2). CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value.