Table 2 Models performance in prediction of metabolic syndrome based on non-invasive features
Data status | Accuracy | Sensitivity | Specificity | PPV | NPV | f1-Score | ROC | Model type & parameters | |
|---|---|---|---|---|---|---|---|---|---|
Imbalance data | Men | 86.9 | 37.1 | 96.2 | 65.2 | 89.1 | 47.2 | 0.86 | J = 200, m = 6 |
Women | 79.4 | 38.2 | 93.4 | 66.7 | 81.7 | 48.3 | 0.79 | J = 200, m = 5 | |
total | 83.6 | 35.9 | 95.3 | 65.3 | 85.8 | 46.3 | 0.83 | J = 200, m = 6 | |
Balancing data based on SMOTE | men | 79.1 | 78.1 | 79.2 | 41.5 | 95.1 | 54.2 | 0.86 | J = 200, m = 4 |
women | 67.6 | 73.4 | 65.7 | 42.2 | 88.0 | 53.5 | 0.79 | J = 200, m = 4 | |
total | 72.1 | 77.6 | 70.7 | 39.5 | 92.8 | 52.3 | 0.83 | J = 200, m = 4 | |
Balancing data based on SplitBal | men | 79.1 | 82.3 | 78.3 | 41.7 | 96.0 | 55.3 | 0.86 | J = 200, m = 5 |
women | 68.1 | 73.7 | 66.1 | 42.6 | 88.1 | 54.0 | 0.80 | J = 200, m = 4 | |
total | 74.2 | 76.3 | 73.6 | 41.7 | 92.7 | 53.9 | 0.83 | J = 200, m = 4 | |