Table 3. Multivariate Cox model performance for the progression outcome under different data partition settings.

The two settings include (a) Using NEPTUNE as training dataset, and CureGN as validation dataset; (b) Using 80% NEPTUNE+CureGN as training dataset and remaining as validation dataset.

Data split settings (training : validation)	Training performance (C-index)	Validation performance (C-index)
100% NEPTUNE: 100% CureGN	0.864±0.030	0.736±0.072
80% (NEPTUNE+CureGN) : 20% (NEPTUNE+CureGN)	0.822±0.015	0.757±0.071