Skip to main content

Table 3 Evaluation results of each model using test set B

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 regression31.54% [31.50,31.58%]94.52% [94.48,94.56%]47.30% [47.25,47.35%]71.85% [71.82,71.88%]
Naïve Bayesian41.98% [41.92,42.04%]83.44% [83.40,83.48%]55.86% [55.80,55.92%]82.37% [82.33,82.41%]
Bayesian network42.95% [42.91,42.99%]84.12% [84.08,84.16%]56.87% [56.82,56.91%]83.06% [83.04,83.08%]
Decision tree(C4.5)33.18% [33.14,33.22%]95.55% [95.51,95.59%]49.26% [49.21,49.31%]71.15% [71.12,71.18%]
Neural network32.72% [32.69,32.75%]94.86% [94.84,94.88%]48.66% [48.62,48.69%]80.33% [80.31,80.35%]
Random forest51.34% [51.31,51.37%]92.81% [92.78,92.84%]66.11% [66.08,66.14%]82.52% [82.49,82.55%]
Bagging with C4.5 decision tree33.06% [33.04,33.08%]94.57% [94.52,94.62%]48.99% [48.96,49.02%]71.02% [70.98,71.06%]
Voting39.66% [39.62,39.70%]91.08% [91.03,91.13%]55.26% [55.21,55.31%]85.13% [85.10,85.16%]
Boosting with C4.5 decision tree36.35% [36.30,36.40%]95.82% [95.79,95.85%]52.71% [52.65,52.76%]80.27% [80.25,80.29%]
  1. *The bold in tables is the maximum value of that evaluation standard