Fig. 5
From: A new risk assessment model of venous thromboembolism by considering fuzzy population

Population clustering and feature importance analysis on inpatients from PUMCH. The clustering analysis with 13 VTE risk factors and 3 Padua-score-related features (Padua score, the number of VTE risk factors, and Padua high risk) was conducted on 3284 inpatients from PUMCH including 230 VTE and 3054 non-VTE patients. Each row represented a patient with (labeled with dark green color) or without (labeled with light green color) VTE in this heatmap. Features listed in the columns were labeled dark blue color as a lower value and red color as a higher value. Besides, feature importance of 13 VTE risk factors was evaluated with the SHAP summary plot