Table 2. Performance of each prediction... | American Society of Hematology

Table 2.

Performance of each prediction model according to C-index in the validation cohort

Risk category	GRFS	OS	Relapse	NRM	aGVHD	cGVHD
Cox-PH	0.668	0.740	0.770	0.664	0.651	0.564
Fine-Gray competing risk model	NA	NA	0.719	0.577	0.582	0.516
Random Survival Forest	0.660	0.745	0.788	0.761	0.580	0.577
XGBoost	0.602	0.712	0.756	0.543	0.540	0.573
Gradient Boosting	0.630	0.602	0.754	0.453	0.590	0.505
Component-wise Gradient Boosting	0.663	0.652	0.774	0.585	0.464	0.570
Dynamic DeepHit	0.646	0.710	0.730	0.691	0.537	0.555
Stacked Ensemble Model	0.670	0.763	0.793	0.777	0.656	0.583

Risk category	GRFS	OS	Relapse	NRM	aGVHD	cGVHD
Cox-PH	0.668	0.740	0.770	0.664	0.651	0.564
Fine-Gray competing risk model	NA	NA	0.719	0.577	0.582	0.516
Random Survival Forest	0.660	0.745	0.788	0.761	0.580	0.577
XGBoost	0.602	0.712	0.756	0.543	0.540	0.573
Gradient Boosting	0.630	0.602	0.754	0.453	0.590	0.505
Component-wise Gradient Boosting	0.663	0.652	0.774	0.585	0.464	0.570
Dynamic DeepHit	0.646	0.710	0.730	0.691	0.537	0.555
Stacked Ensemble Model	0.670	0.763	0.793	0.777	0.656	0.583

aGVHD, grade II-IV acute GVHD; cGVHD, chronic GVHD; NA, not applicable.