Table V: Training AUC with and without CPR with heartrate included as a feature from Machine learning and feature engineering for predicting pulse presence during chest compressions

Comparison of Training AUC values and 95% confidence intervals for various models with and without heartrate (HR), where LDA : Linear Discriminant Analysis, QDA :Quadratic Discriminant Analysis, SVM: Support Vector Machine, GMM: Gaussian Mixture Model, LR: Logistic Regression, RF: Random Forest, NN: Neural Network, ConvNN: Convolutional Neural Network.

不同模型在有无心率（heartrate，HR）特征时的训练AUC值及95%置信区间对比，其中LDA：线性判别分析（Linear Discriminant Analysis），QDA：二次判别分析（Quadratic Discriminant Analysis），SVM：支持向量机（Support Vector Machine），GMM：高斯混合模型（Gaussian Mixture Model），LR：逻辑回归（Logistic Regression），RF：随机森林（Random Forest），NN：神经网络（Neural Network），ConvNN：卷积神经网络（Convolutional Neural Network）。