TABLE 2.
The c-index values, mean and SD (range from 0 to 100) from 200 replications for the DNN, RSF, and LASSO models with weak signals under five scenarios: linear effects (scenario 1) and linear effects together with nonlinear effects (scenario 2) or with interactions (scenario 3) or with nonlinear and interaction effects (scenario 4) or with interaction and indicator effects (scenario 5)
| p | DNN | RSF | LASSO | |
|---|---|---|---|---|
| Scenario 1 | 20 | 66.8 (1.3) | 55.5 (5.8) | 67.0 (1.4) |
| 50 | 73.7 (1.3) | 58.2 (7.6) | 74.0 (1.2) | |
| 100 | 78.2 (1.2) | 60.8 (7.6) | 78.6 (1.1) | |
| 500 | 82.1 (1.4) | 62.9 (4.8) | 75.9 (1.5) | |
| Scenario 2 | 20 | 64.6 (1.6) | 53.7 (4.9) | 63.2 (1.4) |
| 50 | 71.3 (1.3) | 57.1 (7.3) | 71.8 (1.2) | |
| 100 | 76.6 (1.2) | 60.1 (7.3) | 76.9 (1.1) | |
| 500 | 81.5 (1.3) | 62.5 (5.0) | 75.9 (1.5) | |
| Scenario 3 | 20 | 67.4 (1.3) | 55.2 (5.9) | 67.5 (1.3) |
| 50 | 73.2 (1.2) | 56.6 (7.7) | 73.6 (1.2) | |
| 100 | 77.7 (1.2) | 60.0 (7.8) | 78.2 (1.1) | |
| 500 | 81.8 (1.4) | 62.7 (5.0) | 75.6 (1.5) | |
| Scenario 4 | 20 | 65.5 (1.5) | 53.5 (5.1) | 63.9 (1.4) |
| 50 | 71.0 (1.3) | 56.3 (7.5) | 71.4 (1.2) | |
| 100 | 76.0 (1.2) | 59.3 (8.0) | 76.5 (1.2) | |
| 500 | 81.3 (1.4) | 62.4 (5.1) | 75.6 (1.5) | |
| Scenario 5 | 20 | 64.8 (1.3) | 54.3 (4.8) | 64.8 (1.4) |
| 50 | 72.0 (1.2) | 56.5 (7.6) | 72.4 (1.2) | |
| 100 | 77.1 (1.2) | 59.6 (7.9) | 77.6 (1.2) | |
| 500 | 81.8 (1.4) | 62.7 (5.1) | 75.1 (1.5) |
Note: The number of predictors is set at p = 20, 50, 100, 500.
Abbreviations: DNN, deep neural network; RSF, random survival forest.