Skip to main content

Table 2 Evaluation results of each model using test set A

From: Using machine learning models to improve stroke risk level classification methods of China national stroke screening

Learning methodPrecision (95% CI)Recall (95% CI)F1-score (95%CI)AUC (95% CI)
Logistic regression91.84% [91.81,91.87%]97.82% [97.76,97.88%]94.74% [94.69,94.78%]99.14% [99.09,99.19%]
Naïve Bayesian69.48% [69.42,69.54%]97.35% [97.31,97.39%]81.09% [81.03,81.14%]98.44% [98.42,98.46%]
Bayesian network69.66% [69.62,69.70%]97.55% [97.53,97.57%]81.28% [81.24,81.31%]98.41% [98.38,98.44%]
Decision tree(C4.5)92.25% [92.21,92.29%]99.83% [99.78,99.88%]95.89% [95.85,95.94%]99.92% [99.90,99.94%]
Neural network92.19% [92.14,92.24%]99.72% [99.68,99.76%]95.81% [95.76,95.85%]99.15% [99.11,99.19%]
Random forest97.33% [97.30,97.36%]98.44% [98.41,98.47%]97.88% [97.85,97.91%]99.94% [99.92,99.96%]
Bagging with C4.5 decision tree92.25% [92.22,92.28%]99.74% [99.71,99.77%]95.85% [95.82,95.88%]99.93% [99.92,99.94%]
Voting94.34% [94.32,94.36%]99.66% [99.63,99.69%]96.93% [96.91,96.95%]99.94% [99.92,99.96%]
Boosting with C4.5 decision tree95.51% [95.48,95.54%]99.92% [99.89,99.95%]97.67% [97.64,97.70%]99.94% [99.91,99.97%]
  1. *The bold in tables is the maximum value of that evaluation standard