Table 4. Comparisons among ToPs/R, existing clinical risk scores, regression methods, and machine learning benchmarks for post-transplantation survival prediction using C-index and AUC (at horizons of 3-months, 1-year, and 3-years).
Train on 2005-2009; predict on 2010-2015.
| Methods | AUC (Mean ± Std) | C-index (Mean ± Std) | ||
|---|---|---|---|---|
| 3-month | 1-year | 3-year | ||
| ToPs/R | .688 ± .001 | .651 ± .001 | .639 ± .009 | .625 ± .007 |
| DRI | .551 ± .014 | .559 ± .016 | .546 ± .014 | .542 ± .013 |
| IMPACT | .598 ± .013 | .593 ± .001 | .585 ± .011 | .574 ± .009 |
| RSS | .593 ± .017 | .599 ± .020 | .584 ± .013 | .580 ± .012 |
| Cox | .588 ± .012 | .581 ± .009 | .560 ± .010 | .565 ± .007 |
| Linear P | .666 ± .018 | .632 ± .009 | .600 ± .008 | .608 ± .008 |
| Logit R | .662 ± .009 | .633 ± .007 | .604 ± .008 | .609 ± .006 |
| AdaBoost | .643 ± .009 | .630 ± .009 | .606 ± .013 | .607 ± .009 |
| DeepBoost | .643 ± .009 | .630 ± .010 | .608 ± .006 | .606 ± .008 |
| LogitBoost | .655± .009 | .632 ± .007 | .602 ± .013 | .607 ± .008 |
| XGBoost | .574 ± .010 | .567 ± .011 | .554 ± .010 | .555 ± .008 |
| DT | .603 ± .009 | .619 ± .008 | .575 ± .009 | .585 ± .007 |
| RF | .641 ± .009 | .628 ± .006 | .613 ± .008 | .606 ± .006 |
| NN | .648 ± .014 | .628 ± .009 | .600 ± .011 | .604 ± .007 |