Table 6 The voting classifier’s performance on two different top 4 feature sets for female MHD patients
From: Use machine learning to help identify possible sarcopenia cases in maintenance hemodialysis patients
Metrics | Top 4 features (P < 0.05) | Top 4 features (P < 0.05, post-CRE/CysC) |
|---|---|---|
ACCTRSa | 66.73% ± 4.28% | 66.54% ± 3.87% |
ACCTESb | 74.29% ± 8.57% | 71.43% ± 10.43% |
AVADc | 12.72% ± 5.77% | 11.38% ± 7.08% |
Precision | 81.86% ± 7.58% | 79.35% ± 7.82% |
Sensitivity | 76.15% ± 13.95% | 72.31% ± 13.85% |
Specificity | 71.25% ± 15.86% | 70.00% ± 10.00% |
F1 Score | 78.04% ± 8.85% | 75.25% ± 10.40% |
AUC | 77.69% ± 7.92% | 78.37% ± 9.11% |