Table 2. The RSF Model and Conventional Cox Regression Model for Predicting Overall Survival in Patients After Lung Transplantation.
| Models | Time of prediction | iAUC/tAUC (95% CI) | P valuea | iBS/PE (95% CI) | P valuea |
|---|---|---|---|---|---|
| RSF model | 1 to 48 mo | 0.879 (0.832-0.921) | [Reference] | 0.130 (0.106-0.154) | [Reference] |
| Cox model | 1 to 48 mo | 0.658 (0.572-0.747) | <.001 | 0.205 (0.176-0.233) | <.001 |
| RSF model | 1 mo | 0.858 (0.792-0.917) | [Reference] | 0.123 (0.096-0.153) | [Reference] |
| Cox model | 1 mo | 0.624 (0.523-0.728) | <.001 | 0.181 (0.100-0.219) | <.001 |
| RSF model | 1 y | 0.921 (0.877-0.957) | [Reference] | 0.115 (0.095-0.139) | [Reference] |
| Cox model | 1 y | 0.717 (0.633-0.800) | <.001 | 0.195 (0.098-0.225) | <.001 |
Abbreviations: Cox, Cox regression; iAUC, integrated area under the curve; iBS, integrated Brier score; PE, prediction error; tAUC, time-dependent area under the curve; RSF, random survival forests.
a
Comparison with the performance of Cox model to RSF model with the same time of prediction.