Fig. 2

Multi-parameter modeling of PAH survival with Elastic Net Cox regression. The Elastic Net multi-parameter Cox regression model with the set of 19 (Supplementary Table S1) independent variables and overall survival as a response was developed in the training Innsbruck cohort. Numeric independent variables were median centered and their first and second order terms included in the model. Numbers of complete observations and mortality is indicated in B. A Non-zero Elastic Net model coefficients (Elastic Net signature) represented as hazard ratios (HR). Plot points are labeled with their HR values. B Association of overall survival with the Elastic Net model linear prediction score in the training IBK and test Linz/Vienna (LZ/W) cohort was assessed by Kaplan–Meier analysis. Significance of the survival differences in the study participants stratified by the linear predictor score tertiles (T1: 0—33, T2: 34—66, T3: 66—100 percentile) was determined by log-rank test adjusted for multiple testing with Benjamini–Hochberg method. P values are shown in the plots, numbers of complete observations and mortality are indicated in the plot captions