基于机器学习方法的慢性阻塞性肺疾病分期预测

Stage Prediction of Chronic Obstructive Pneumonia Based on Machine Learning

目的:采用基于机器学习的分类判断算法,建立慢性阻塞性肺疾病(以下简称"慢阻肺")分期模型,提高慢阻肺诊断和分期的准确度。方法:选择确诊慢阻肺住院患者2 504例,以国际GOLD分期为依据,收集与慢阻肺分期密切相关的临床特征参数指标,对参数进行筛选,参照医院的临床确诊结果,采用机器学习方法(k-最近相邻法、SVM)训练并测试慢阻肺的分期模型。结果与结论:数据为不平衡数据,虽采用分层比例抽样,但针对此类数据SVM的准确率更高为85.26%。说明机器学习提供的模型能为慢阻肺分期提供较准确的分类依据。

Objective:To improve the accuracy of diagnosis and staging of chronic obstructive pulmonary disease by using classification and judgment algorithm based on machine learning.Methods:2504 hospitalized patients with COPD were selected.According to the international GOLD staging,the clinical characteristic parameters closely related to COPD staging were collected,and the parameters were screened.Referring to the clinical diagnosis results of the hospital,the machine learning method(k-nearest neighbor method,SVM)was used to train and test the staging model of COPD.Results and Conclusion:The data are unbalanced.Although stratified proportional sampling is used,the accuracy of SVM for such data is higher 85.26%.It shows that the model provided by machine learning can provide a more accurate classification basis for COPD staging.