Table 1. Performance of amplitude spectrum area (AMSA) and three different combination strategies for the prediction of subsequent shocks in validation data (51 successful and 122 failed shocks) by the BP neural network.
| AUC | Sensitivity(%) | Specificity(%) | NPV(%) | PPV(%) | PA(%) | |
|---|---|---|---|---|---|---|
| AMSA | 0.819 | 35.3 | 90.2 | 76.9 | 60.0 | 74.0 |
| C1 | 0.883** | 68.6## | 91.0 | 87.4# | 76.1 | 84.4# |
| C2 | 0.825 | 49.0 | 90.2 | 80.9 | 67.6 | 78.0 |
| C3 | 0.904** | 76.5## | 90.2 | 90.2## | 76.5 | 86.1## |
C1: AMSA+PSI;C2: AMSA+ΔAMSA; C3: AMSA+PSI+ΔAMSA;PSI: previous shock index; AUC: area under receiver operating characteristic curve; NPV: negative predictive value; PPV: positive predictive value; PA: prediction accuracy.
**p<0.01 compared with AMSA using Z-test;
#, ##p<0.05 and p<0.01 compared with AMSA using Chi-squared test.