基于机器学习与心肺复苏诊疗标准的辅助诊疗算法

为了将机器学习在心肺复苏领域的应用落实到现实临床工作之中，本研究提出了一种基于机器学习并融合心肺复苏诊疗标准的辅助诊疗算法。采用了随机森林、梯度提升树、极端梯度提升作为基模型，使用投票法进行模型融合，并引入模型解释算法(Shapley Additive explanation，SHAP)过滤掉Shapley值较低的特征重新进行训练，得出的模型在心肺复苏诊疗标准下创建参数空间进行寻优，最终得到最优诊疗方案。结果表明，融合心肺复苏诊疗标准后的算法更符合临床实际，可为临床诊疗提供辅助，提高心肺复苏成功率。

Research on auxiliary diagnosis and treatment algorithm based on machine learning and cardiopulmonary resuscitation diagnosis and treatment standards

In order to implement the application of machine learning in the field of cardiopulmonary resuscitation (CPR) into real clinical work, this study proposes an assisted diagnosis and treatment algorithm based on machine learning and fused with CPR diagnosis and treatment standards. Random Forest, Gradient Boosting Tree and Extreme Gradient Boosting are used as the base model. The voting method is used for model fusion, and the Shapley Additive explanation algorithm (SHAP) is introduced to filter out the features with lower Shapley values for retraining, and the resulting model creates a parameter space under the cardiopulmonary resuscitation diagnostic and treatment standard for optimization, and finally obtains the optimal diagnostic and treatment plan. The resulting model was optimized by creating a parameter space under the CPR diagnosis and treatment criteria, and the optimal treatment plan was finally obtained. The results show that the algorithm after integrating the CPR diagnosis and treatment standards is more in line with the clinical reality, which can provide assistance for clinical diagnosis and treatment and improve the success rate of CPR.