Skip to main content
Fig. 3 | BMC Medical Research Methodology

Fig. 3

From: Multivariate longitudinal data for survival analysis of cardiovascular event prediction in young adults: insights from a comparative explainable study

Fig. 3

Explanation for the RSF model trained on trajectory clustering data. Top-left panel: normalized median variable importance (VIMP) over 10 folds from permutation for the input variables (trajectory membership and demographic variables) of RSF on trajectory clustering data. Top-right panel: cluster profiles for each longitudinal variable, showing the representative (median) trajectory per cluster. The clustered plots are ordered by the ranking on the left. Bottom panel: partial dependence plots showing the effects of cluster membership (x-axis) on survival probability on the y-axis (1-CVD risk). The colors correspond to the cluster median trajectories in the top panel. DBP: diastolic blood pressure, HBM: taking hypertensive medication (yes/no), GLU: glucose, SMKNW: smoking regularly now (yes/no), KIDNY: kidney problem (yes/no), LDL: low-density lipoprotein, SBP: systolic blood pressure, WGT: weight. The full variable names are explained in the abbreviation section

Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us