Kokiri: Random Forest-Based Cohort Comparison and Characterization

We propose an interactive visual analytics approach for the characterization and comparison of patient subgroups (i.e., cohorts). Despite having the same disease and similar demographic characteristics, patients respond differently to therapy. One reason for this is the vast number of variables in the genome that influence the outcome. Nevertheless, most existing tools do not offer effective means to identify the most differing attributes or look at them in isolation, neglecting combinatorial effects. To fill this gap, we present Kokiri, a visual analytics approach that aims to separate cohorts based on user-selected data, ranks attributes by their importance to distinguish between cohorts, and visualizes the overlap and separability of the cohorts. Using our approach, users can additionally characterize the homogeneity and outliers of the cohort. To demonstrate the applicability of our approach, we integrated Kokiri in the Coral cohort analysis tool for the purpose of comparing and characterizing lung cancer patient cohorts. ### Competing Interest Statement The authors have declared no competing interest.