基于机器学习的新冠肺炎典型药物疗效分析

Analysis of the efficacy of typical drugs for COVID-19based on machine learning

目的:新型冠状病毒肺炎具有传播能力强、检测准确率低、治疗难度大等特点,对我国乃至全世界人类健康和社会安全造成了巨大威胁。新冠肺炎目前没有特效药物,因此,需要比较各类药物对新冠肺炎的治疗效果,为重症肺炎患者的治疗提供参考。方法:本文选取某院1384名重症出院患者作为数据集,提取最广泛使用的三类药物作为训练特征,构建基于药物使用情况预测患者治愈率的多个机器学习模型,最后基于综合性能最优的模型,分析三类药物对于治愈率的重要性。结果:年龄是影响治愈率的最重要因素;激素类药物在所有药物中影响占比最大,中成药的疗效也非常突出。结论:相对于传统的针对单个药物药效的统计学研究方法,机器学习可以将所有药物药效进行综合分析,能直观地显示各个药物对治愈率的影响程度。

OBJECTIVE COVID-19(Corona Virus Disease 2019)has characteristics of strong transmission ability,low detection accuracy,and difficult treatment,which had posed huge threat to mankind health and social security in China and all over the world.There is no specific drug for COVID-19.Therefore,it is necessary to compare the therapeutic effects of various drugs on new-coronary pneumonia and provide a reference for the treatment of patients with severe pneumonia.METHODS A total of 1384critically ill patients discharged from Tongji Hospital were selected as the data set,and the three most widely used drugs were extracted as the training characteristics to construct multiple machine learning models for predicting the cure rate of patients based on drug use.Finally,based on the model with optimal comprehensive performance,the importance of the three drugs for the cure rate was analyzed.RESULTS Age was the most important factor affecting the cure rate;hormone drugs accounted for the largest proportion of all drugs,and the efficacy of Chinese patent medicines was also very significant.CONCLUSION Compared with the traditional statistical research method for the efficacy of a single drug,machine learning can comprehensively analyze the efficacy of all drugs and can visually show the degree of influence of each drug on the cure rate.