北京生物医学工程

基于实验室指标的新型冠状病毒肺炎鉴别诊断模型

Diagnosis model of COVID-19 based on laboratory indicators

作者：朱碧云王妮陈卉应晓飞康娜张淳

单位：首都医科大学附属北京地坛医院医学工程处（北京100015)<br />首都医科大学生物医学工程学院（北京 100069)<br />通信作者：张淳。E-mail：991716945@qq.com

分类号：R318. 04

出版年·卷·期（页码）：2022·41·5（483-487）

摘要：

目的基于实验室指标数据建立新型冠状病毒肺炎与流感病毒性肺炎的鉴别诊断模型，并评价模型的性能，为两种疾病的鉴别诊断提供依据。方法收集2020年1月至6月住院的175名新型冠状病毒肺炎患者和2019年同期住院的157名流感病毒性肺炎患者入院后的首次实验室数据，分别利用机器学习中的决策树及决策树的集成算法随机森林和XGBoost（eXtreme gradient boosting）建立鉴别诊断模型，通过准确率、F1分数和接受者操作特征（receiver operating characteristic, ROC）曲线下面积评价和比较三种模型的预测效果。结果决策树、随机森林和XGBoost模型的准确率分别为0.831、0.892和0.898， F1分数分别为0.836、0.894和0.902，ROC曲线下的面积分别为0.862、0.958和0.963。随机森林和XGBoost模型的诊断性能明显优于决策树模型。结论利用实验室指标能够建立高性能的新型冠状病毒肺炎鉴别诊断机器学习模型，这些模型有望帮助医生进行新型冠状病毒肺炎与流感病毒性肺炎的鉴别诊断。

Objective The differential diagnosis models of corona virus disease 2019(COVID-19) and influenza virus pneumonia are established using laboratory test items ，the performance of the models are evaluated to provide a basis for the differential diagnosis of the two diseases. Methods First time laboratory data of 175 cases of COVID-19 between January and June 2020 and 157 patients hospitalized for influenza virus pneumonia during the same period in 2019 were collected and the decision tree in machine learning and its integrated algorithms random forest and XGBoost (eXtreme Gradient Boosting) were used to build differential diagnosis models. The prediction effects of the three models were compared by the accuracy rate, F1 score and the area under the receiver operating characteristic (ROC) curve. Results The accuracy rates of three models were 0.831, 0.892 and 0.898, the F1 scores were 0.836, 0.894 and 0.902, and the areas under ROC curves were 0.862, 0.958 and 0.963, respectively. The performances of random forest and XGBoost model were better than that of decision tree model. Conclusions High-performance COVID-19 differential diagnosis machine models can be built using laboratory indicators, which are expected to help doctors differentiate COVID-19 from influenza virus pneumonia.

参考文献：

[1] Worldometer. COVID-19 coronavirus pandemic [EB/OL]. [2021-01-19]. http://www.worldometers.info/coronavirus/#countries.
[2] 白君, 周洁, 马婉玲, 等. 新型冠状病毒肺炎的初诊CT特征及与流感病毒性肺炎的鉴别诊断[J]. 中国中西医结合影像学杂志, 2020, 18(3):246-248.
Bai J, Zhou J, Ma WL, et al. CT Features of COVID-19 at initial diagnosis and comparing with those of influenza virus pneumonia[J]. Chinese Imaging Journal of Integrated Traditional and Western Medicine, 2020, 18(3):246-248.?
[3] 兰长利. 不同类型新型冠状病毒肺炎的肺脏超声的动态表现[J].中国医药导报，2021,18(10)：151-154,158.
Lan CL.Dynamic manifestations of lung ultrasound with different types of corona virus disease 2019[J]. China Medical Herald，2021,18(10)：151-154,158.
[4] 白欢，沈玲，袁旭，等.新型冠状病毒肺炎患者早期外周血实验室检查结果分析[J].现代检验医学杂志,2020,35(5):93-98.
Bai H, Shen L, Yuan X, et al. Analysis of the characteristics of hematological parameters in the early stages of COVID-19[J]. Journal of Modern Laboratory Medicine, 2020,35(5):93-98.
[5] 刘子砚,牛志杰,井沆,等. 新型冠状病毒肺炎患者347例实验室检测结果分析[J]. 现代医药卫生,2021,37(4):572-575.
Liu ZY, Niu ZJ, Jing H, et al. Analysis of laboratory test results of 347 patients with COVID-19[J]. Journal of Modern Medicine & Health, 2021,37(4):572-575.
[6] 曾雪元, 宫伟国, 胡云峰, 等. 基于决策树算法构建缺血性卒中复发的预测模型[J]. 吉林中医药, 2020, 40(4):437-440.
Zeng XY, Gong WG, Hu YF, et al. A model for predicting ischemic stroke recurrence based on decision tree algorithm[J]. Jilin Journal of Chinese Medicine, 2020, 40(4):437-440.?
[7] 杨青, 王国蓉, 江宾, 等. 基于决策树的肿瘤患者难免性压疮风险预测模型研究[J]. 护理学杂志, 2019, 34(13):4-7.
Yang Q, Wang GR, Jiang B, et al. Study on risk prediction model of unavoidable pressure ulcers in cancer patients based on decision tree[J]. Journal of Nursing Science, 2019, 34(13):4-7.
[8] 于大海, 李金, 罗艳虹, 等. 随机森林模型和决策树模型在肝硬化上消化道出血预后中的应用[J]. 中国卫生统计, 2019, 36(2):162-166.
Yu DH, Li J, Luo YH. Application of random forest and decision trees in the prognosis of upper gastrointestinal bleeding in patients with liver cirrhosis[J]. Chinese Journal of Health Statistics, 2019, 36(2):162-166.
[9] 吴懿婷.决策树和随机森林方法在管理决策中的应用[J].电子制作, 2018,(16):47-48,100.
Wu YT. Application of decision tree and random forest method in management decision[J]. Practical Electronics, 2018,(16):47-48,100.?
[10] 陈卓, 吕娜. 基于随机森林和XGBoost的网络入侵检测模型[J]. 信号处理, 2020, 36(7):1055-1064.
Chen Z, LYu N. Network intrusion detection model based on random forest and XGBoost[J]. Journal of Signal Processing, 2020, 36(7):1055-1064. DOI:10.16798/j.issn.1003-0530.2020.07.004.
[11] 谷鸿秋, 王春娟, 李子孝, 等. 基于Logistic回归与XGBoost构建缺血性卒中院内复发风险预测模型的初步比较研究[J]. 中国卒中杂志, 2020, 15(6):587-594.
Gu HQ, Wang CJ, Li ZX, et al. Comparison of prediction models for in-hospital stroke recurrence in patients with ischemic stroke based on logistic regression and XGBoost methods[J]. Chinese Journal of Stroke, 2020, 15(6):587-594.
[12] 袁宏丽,范恒全,马彦华.新型冠状病毒肺炎与其他常见病毒性肺炎的CT鉴别诊断[J]. 解放军医药杂志,2020,32(4):5-8.
Yuan HL, Fan HQ, Ma YH. CT differential diagnosis between novel coronavirus and other common viral pneumonia[J]. Medical & Pharmaceutical Journal of Chinese People's Liberation Army, 2020,32(4):5-8.
[13] 蒋荣猛. 新型冠状病毒肺炎诊疗方案(试行第七版)解读[J]. 北京医学, 2020, 42(4):334-336.
Jiang RM. “New coronavirus pneumonia treatment plan(trial seventh edition)”interpretation[J]. Beijing Medical Journal, 2020, 42(4):334-336.
[14]吴敏昱, 李敏, 朱芸仙, 等. 新型冠状病毒肺炎与甲型H1N1流感肺炎CT特征比较[J]. 中国医学影像学杂志, 2020, 28(11):802-805.?
Wu MY, Li M, Zhu YX, et al. CT features of COVID-19 and H1N1 influenza pneumonia: a comparative study[J]. Chinese Journal of Medical Imaging, 2020, 28(11):802-805.
[15] 高瞻，陈文，姜鹏，等.基于XGBoost的新冠肺炎智能检测系统实现[J].中国数字医学, 2020,15(11):123-127.
Gao Z, Chen W, Jiang P, et al. Realization of the intelligent detection system for COVID-19 based on XGBoost[J]. China Digital Medicine, 2020, 15(11):123-127.

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录