Table 3 The voting classifier’s evaluation metrics about four feature sets in male MHD patients
From: Use machine learning to help identify possible sarcopenia cases in maintenance hemodialysis patients
Metric | Top 3 Features | Top 4 Features | Top 5 Features | Top 5 Features (P < 0.05) |
|---|---|---|---|---|
ACCTRSa | 86.59% ± 1.89% | 73.03% ± 1.41% | 78.48% ± 2.82% | 74.92% ± 2.41% |
ACCTESb | 80.71% ± 4.29% | 75.36% ± 4.06% | 79.64% ± 6.79% | 75.00% ± 4.79% |
AVADc | 6.61% ± 3.63% | 4.88% ± 2.70% | 8.00% ± 3.02% | 3.69% ± 3.45% |
Precision | 79.28% ± 9.86% | 70.65% ± 6.79% | 77.90% ± 13.17% | 70.61% ± 10.71% |
Sensitivity | 77.50% ± 11.21% | 75.83% ± 10.83% | 78.33% ± 13.54% | 75.83% ± 6.92% |
Specificity | 83.12% ± 9.70% | 75.00% ± 10.08% | 80.62% ± 12.95% | 74.38% ± 10.63% |
F1 Score | 77.32% ± 5.36% | 72.28% ± 5.12% | 76.58% ± 7.61% | 72.36% ± 4.11% |
AUC | 87.40% ± 4.41% | 85.57% ± 3.86% | 86.04% ± 5.35% | 85.05% ± 4.76% |