Survival prediction models: an introduction to discrete-time modeling

Table 3 Cross-validated R-squared Brier score for continuous and discrete-time survival prediction models in multiple data sets

Method	metabric	flchain	nwtco	colon	pbc
Cox PH	0.140	0.351	0.112	0.111	0.334
RSF	0.133	0.344	0.140	0.128	0.389
Logistic regression	0.136	0.352	0.114	0.112	0.328
Elastic net	0.136	0.350	0.116	0.114	0.353
Support vector machine	0.070	0.101	0.018	0.046	0.349
GBM	0.128	0.364	0.149	0.121	0.378
Neural network	0.163	0.366	0.136	0.116	0.415
CForest	0.138	0.343	0.147	0.123	0.377

Higher values indicated better predictive performance. Bold values indicate method with best predictive performance (highest R-squared) in a particular data set. CForest: conditional inference random forest; GBM: gradient boosting machines; PH: proportional hazards; RSF: random survival forest

ISSN: 1471-2288