Skip to main content
Fig. 1 | Radiation Oncology

Fig. 1

From: Single-center versus multi-center data sets for molecular prognostic modeling: a simulation study

Fig. 1

Simulation scheme for the computation of performance scores of molecular prognostic models with center effects. For each parameter set 1000 Monte-Carlo runs are performed. In single-center (SC) and multi-center (MC) data sets, independent variables X and dependent variables Y are generated from a true state a, noise and the center batch pattern. Each center shares realizations of randomly sampled batch-specific parameters among its samples. MC data sets are batch corrected, a signature is fitted to the discovery data and used for prediction of validation data. Performance scores are calculated to measure the average quality of prediction

Back to article page