Figure 6
From: Predicting CKD progression using time-series clustering and light gradient boosting machines
Importance matrix plot of the LightGBM model in Class 4 vs. Class 5 patients, representing the importance of each covariate (Model 3) for predicting classification of 5-year eGFR trajectory (a). SHAP summary plot of clinical features of the LightGBM model in Class 4 and 5 patients. One dot represents a patient’s colored feature value, where red has a higher value and blue has a lower value (b). Model 3 adjusted for age, sex, body mass index, smoking history, systolic blood pressure, diastolic blood pressure, history of cardiovascular disease, diabetes, eGFR, LDL-cholesterol, serum albumin, hemoglobin, platelet count, MCHC, RDW, qualitative test for proteinuria, use of an ACE inhibitor or ARB, and use of a xanthine oxidase inhibitor. GBM gradient boosting machine, eGFR estimated glomerular filtration rate, LDL low-density lipoprotein, MCHC mean corpuscular hemoglobin, RDW red cell distribution width, ACE angiotensin-converting enzyme, ARB angiotensin II receptor blocker.
