Prehospital diagnostic algorithm for acute coronary syndrome using machine learning: a prospective observational study

Figures

• 

  Figure 1. Relationship between the number of features and the area under the receiver operating characteristic curve for the prediction algorithm. The line plot depicts sequential changes in the AUC with the number of features for the prediction algorithm in (a) the training score (blue) and (b) the test score (yellow). The dotted vertical line indicates the highest predictive value in the test score. (n = 17, AUC of the training score = 0.881, AUC of the test score = 0.859). The error bars indicate 95% confidence intervals. AUC (area under the receiver operating characteristic curve).

• 

  Figure 2. Receiver operating characteristic curve of prehospital diagnostic algorithms for acute coronary syndrome with 17 features. ROC curves of the top six machine learning algorithms for the prehospital prediction of ACS using 17 features are shown. The ROC curves are depicted at 1-specificity on the x-axis and sensitivity on the y-axis using (a) the training score, (b) the test score, and (c) external cohort score. AUC is presented with 95% confidence interval. ACS (acute coronary syndrome), AUC (area under the receiver operating characteristic curve), CI (confidence interval), LDA (linear discriminant analysis), LR (logistic regression), MLPC (multilayer perceptron classifier), ROC (receiver operating characteristic), SVM (R) (support vector machine radial basis function), VC (voting classifier), XGB (eXtreme Gradient Boosting).

• 

  Figure 3. SHAP values of the prehospital diagnostic algorithm for acute coronary syndrome using 17 features. The impact of the features on the model output was expressed as the SHAP value calculated with the support vector machine (radial basis function). The features are placed in descending order according to their importance. The association between the feature value and SHAP value indicates a positive or negative impact of the predictors. The extent of the value is depicted as red (high) or blue (low) plots. SHAP (SHapley Additive exPlanation).