Fig. 6: DyAM-based multimodal analysis.
a, HRs and single feature AUCs of covariates. Overall risk score analysis: forest plot of the DyAM model score with respect to other clinical variables (left). Partial risk score analysis: forest plot of the modality specific risk scores from LR compared to DyAM (right). The vertical dashed lines represent a null HR. P values were obtained from the likelihood-ratio test. The bar height and error bar represent the AUC and associated 95% CI based on DeLong’s method51 for n = 247 patients. b, A zoom in of the first 12 months separated by quartile showing separation of early progression events. P values were obtained from the log-rank test. c, Alpine plots comparing the overall model score after reweighting the input modalities as a function of a multiplier to a single modality’s attention, for the model’s AUC (left), HR (middle) and PFS ratio at 4 months in the lower and higher quartiles (right). For the reweighted AUCs, HRs and PFS ratios, the error bar represents the 68% or 1-sigma CIs based on DeLong’s method51, the HR covariance matrix and Poisson standard error on the number of progression events, respectively.
