Table 3 Subgroup analyses for prediction of ESKD at 3 years
From: Machine learning models to predict end-stage kidney disease in chronic kidney disease stage 4
Subgroup | AUROC | |||
|---|---|---|---|---|
| Â | LASSO regression | Random forest | XGBoost | Neural network |
eGFR | Â | Â | Â | Â |
15–20 mL/min/1.73m2 (n = 413) | 0.69 (0.63 to 0.74) | 0.69 (0.64 to 0.75) | 0.65 (0.63 to 0.74) | 0.70 (0.65 to 0.75) |
20–29 mL/min/1.7m2 (n = 1,747) | 0.74 (0.72 to 0.77) | 0.74 (0.71 to 0.76) | 0.73 (0.71 to 0.76) | 0.75 (0.72 to 0.77) |
Sex | Â | Â | Â | Â |
Female (n = 1,695) | 0.79 (0.76 to 0.82) | 0.78 (0.75 to 0.81) | 0.77 (0.74 to 0.81) | 0.79 (0.76 to 0.82) |
Male (n = 1,465) | 0.75 (0.71 to 0.78) | 0.74 (0.71 to 0.77) | 0.74 (0.71 to 0.78) | 0.75 (0.72 to 0.78) |
Race | Â | Â | Â | Â |
Black (n = 401) | 0.74 (0.68 to 0.80) | 0.75 (0.69 to 0.81) | 0.75 (0.66 to 0.79) | 0.74 (0.68 to 0.80) |
Hispanic (n = 320) | 0.73 (0.66 to 0.80) | 0.71 (0.64 to 0.79) | 0.71 (0.65 to 0.79) | 0.73 (0.66 to 0.80) |
White (n = 1,211) | 0.77 (0.73 to 0.81) | 0.78 (0.74 to 0.81) | 0.78 (0.72 to 0.81) | 0.77 (0.73 to 0.81) |
Others (n = 1,228) | 0.78 (0.74 to 0.81) | 0.76 (0.73 to 0.80) | 0.76 (0.74 to 0.81) | 0.78 (0.75 to 0.81) |
Comorbidity | Â | Â | Â | Â |
BMI ≥ 25 kg/m2 (n = 1,424) | 0.78 (0.75 to 0.80) | 0.78 (0.75 to 0.80) | 0.77 (0.75 to 0.80) | 0.78 (0.76 to 0.81) |
Diabetes (n = 1,522) | 0.76 (0.73 to 0.79) | 0.71 (0.73 to 0.79) | 0.76 (0.72 to 0.79) | 0.76 (0.72 to 0.79) |
Cardiovascular disease | 0.75 | 0.77 | 0.76 | 0.77 |
(n = 1,447) | (0.72 to 0.79) | (0.73 to 0.80) | (0.72 to 0.79) | (0.73 to 0.80) |
Laboratory value | Â | Â | Â | Â |
K > 5.5 mEq/L | 0.82 | 0.82 | 0.80 | 0.82 |
(n = 282) | (0.75 to 0.89) | (0.75 to 0.89) | (0.74 to 0.88) | (0.75 to 0.89) |
Phosphate > 5Â mg/dL | 0.73 | 0.71 | 0.68 | 0.73 |
(n = 114) | (0.64 to 0.81) | (0.61 to 0.80) | (0.62 to 0.80) | (0.64 to 0.82) |
Serum albumin < 3.0 | 0.80 | 0.79 | 0.78 | 0.78 |
(n = 110) | (0.73 to 0.86) | (0.72 to 0.86) | (0.70 to 0.85) | (0.71 to 0.85) |