Table 2.
Prediction ability of the reference and four machine-learning-based prediction models for HNHC patients.
| Outcome | c-statistics | P-valueb | Sensitivity | Specificity | PPV | NPV | PLR | NLR |
|---|---|---|---|---|---|---|---|---|
| Reference modela | 0.82 (0.81–0.84) | [Reference] | 0.74 (0.72–0.76) | 0.78 (0.77–0.78) | 0.15 (0.14–0.16) | 0.98 (0.98–0.98) | 3.3 (3.2–3.4) | 0.34 (0.31–0.36) |
| Logistic regression with Lasso regularization | 0.82 (0.81–0.84) | 0.98 | 0.67 (0.65–0.79) | 0.85 (0.84–0.85) | 0.19 (0.18–0.20) | 0.98 (0.98–0.98) | 4.3 (4.2–4.5) | 0.39 (0.37–0.42) |
| Random forest | 0.84 (0.83–0.85) | 0.12 | 0.71 (0.69–0.73) | 0.83 (0.83–0.84) | 0.18 (0.17–0.19) | 0.98 (0.98–0.98) | 4.2 (4.1–4.4) | 0.35 (0.32–0.37) |
| Gradient-boosted decision tree | 0.84 (0.83–0.86) | 0.01 | 0.68 (0.66–0.71) | 0.88 (0.87–0.88) | 0.22 (0.21–0.24) | 0.98 (0.98–0.98) | 5.5 (5.3–5.7) | 0.36 (0.34–0.39) |
| Deep neural network | 0.84 (0.83–0.85) | 0.03 | 0.71 (0.68–0.73) | 0.85 (0.85–0.86) | 0.20 (0.19–0.21) | 0.98 (0.98–0.98) | 4.8 (4.6–5.0) | 0.35 (0.32–0.37) |
HNHC high-need, high-cost, PPV positive predictive value, NPV negative predictive value, PLR positive likelihood ratio, NLR negative likelihood ratio.
aWe used a non-penalized logistic regression model as the reference model.
bWe compared the area under the curve between each machine-learning-based prediction model and the logistic regression model (the reference model) using the DeLong’s test.