Table 2 The performance metrics of four machine learning models
Models | tr + v:te | Metric | RIF | INH | EMB | PZA |
|---|---|---|---|---|---|---|
WDNN | 3601:792 | Sens | 95.4% | 90.3% | 90.6% | 75.2% |
Spe | 97.9% | 96.4% | 85.6% | 91.2% | ||
Acc | / | / | / | / | ||
NPV | / | / | / | / | ||
PPV | / | / | / | / | ||
AUC | 98.2% | 95.9% | 92.2% | 88.3% | ||
resis:sus | / | / | / | / | ||
DeepAMR | 7:3 | Sens | 94.2% | 94.3% | 91.5% | 87.3% |
Spe | 95.8% | 95.7% | 93.4% | 90.9% | ||
Acc | / | / | / | / | ||
NPV | / | / | / | / | ||
PPV | / | / | / | / | ||
AUC | 98.2% | 97.7% | 96.8% | 94.4% | ||
resis:sus | / | / | / | / | ||
GBT-CRM | 8:2 | Sens | 88.8% | 91.1% | 82.8% | 69.7% |
Spe | 98.9% | 98.8% | 94.2% | 96.1% | ||
Acc | 96.2% | 96.3% | 92.1% | 91.8% | ||
NPV | 96.0% | 95.8% | 96.1% | 94.2% | ||
PPV | 96.8% | 97.4% | 75.6% | 78.0% | ||
AUC | 97.9% | 96.7% | 95.8% | 95.5% | ||
resis:sus | 4462/12045 | 5215/11207 | 2576/12254 | 1813/10155 | ||
TB-DROP | 7:3 | Sens | 89.9% | 88.3% | 90.4% | 90.7% |
Spe | 90.6% | 90.0% | 84.9% | 81.5% | ||
Acc | 90.4% | 89.5% | 85.8% | 82.6% | ||
NPV | 96.2% | 94.5% | 98.0% | 98.4% | ||
PPV | 77.9% | 80.0% | 52.4% | 41.0% | ||
AUC | 95.4% | 94.6% | 93.2% | 90.5% | ||
resis:sus | 3266/9086 | 3770/8493 | 1869/10172 | 1367/9802 |