ESTRO38 Congress Report

A

C

B

D

Figure 1: Dashboard with results of the distributed data analysis and machine learning algorithms. ( A ) The number of patients available for training or validation per year per site. ( B ) Training progress of the iterative distributed logistic regression (LR) algorithm. Left axis: Root mean square error (RMSE) for training and validation cohorts of LR models optimized on the training cohort at a given iteration. Right axis: regression coefficients for T, N, M, and overall stage categories computed by the LR algorithm at a given iteration. ( C ) Discriminative model performance according to the receiver operating characteristic curve (ROC) and its area under the curve (AUC) on the validation cohort per site. ( D ) Calibration performance on a single site (the site with most data).

Congress report | AWARDS

51