Table 2.
Nested cross-validation results for each prediction model
| Model | Sensitivity | PPV | Specificity | NPV | BAC | Accuracy |
|---|---|---|---|---|---|---|
| AVH – intensity | ||||||
| Physio | 82% | 69% | 81% | 90% | 82% | 81% |
| Physio+PSYRATS | 90% | 75% | 84% | 94% | 87% | 86% |
| AVH – distress | ||||||
| Physio | 84% | 65% | 80% | 92% | 82% | 81% |
| Physio+PSYRATS | 90% | 71% | 84% | 95% | 87% | 86% |
| Paranoia – intensity | ||||||
| Physio | 76% | 64% | 75% | 84% | 76% | 75% |
| Physio+PSYRATS | 87% | 76% | 84% | 92% | 86% | 85% |
| Paranoia – distress | ||||||
| Physio | 75% | 50% | 66% | 85% | 71% | 69% |
| Physio+PSYRATS | 87% | 64% | 77% | 93% | 82% | 80% |
Note: Physio, model based on only physiological data; Physio+PSYRATS, model based on physiological data and PSYRATS scores; PPV, positive predictive value; NPV, negative predictive value; BAC, balanced accuracy.