Fig. 5From: A non-linear optimisation method to extract summary statistics from Kaplan-Meier survival plots using the published P valueAdditional validation of the nlopt method. The KM plots from five additional datasets (a-e) accessible from the R package ‘survival’ were plotted (left panel) and the ln (HR) and var. ln (HR) calculated using the rules established from Figs. 2, 3 and 4. The individual points chosen for each KM plot are overlaid on the KM plot (middle panel) and represent the exact points chosen for each dataset. The object in R used to access each dataset is included as well as the starting number at risk for each arm, the number of points used and the computed and actual values for ln (HR) and var. ln (HR) (right panel). The mean absolute error and mean % absolute error is summarised at the bottom right of the figureBack to article page