Skip to main content
Fig. 1 | BMC Medical Research Methodology

Fig. 1

From: Framework for personalized prediction of treatment response in relapsing remitting multiple sclerosis

Fig. 1

Comparison of predicted therapy. a Patients assigned to the same highest ranked therapy are divided into two groups: those who received indeed this therapy (red) and those who received another therapy (gray). b Weights are calculated for the two groups to mitigate the effect of confounds on the analysis. In particular, the weights are calculated such that each group matches the population statistics. As an example, this results in larger weights for females in the group with a smaller ratio of females compared with the overall occurrence than in the group with a larger ratio. c The weights are included in a survey-weighted GLM, where an indicator variable encodes membership to one of the two groups, and observation time is accounted for. The GLM allows for a propensity-score-based weighting of the clinically relevant outcomes of the two groups. The estimated slope allows then for a comparison of the disease activity between the two groups

Back to article page