摘要
目的使用XGBoost算法建立阻塞性睡眠呼吸暂停(OSA)的机器学习预测模型。方法纳入680例于首都医科大学附属北京同仁医院行整夜多道睡眠图监测患者,收集其性别、年龄、颈围以及体质量指数(body mass index,BMI)、夜间最低血氧饱和度、夜间平均血氧饱和度、3%氧减指数等信息,作为机器学习的输入特征,以XGBoost算法建立OSA的自动预测模型,并与其他几种常用的机器学习算法(如支持向量机、随机森林、决策树等)建立的模型结果进行对比。结果在四分类(正常受试者,轻度、中度、重度OSA)中,XGBoost分类器在所有算法中表现最好,综合分类准确率为93.4%,F1值得分分别为94.7%、78.7%、86.7%、98.1%,ROC曲线下面积(AUC)分别为98.0%、85.0%、92.0%和98.0%。结论基于XGBoost算法成功建立成人OSA预测模型,可用于临床OSA的诊断与筛查。
OBJECTIVE To build a machine learning model that could automatically diagnose obstructive sleep apnea(OSA) using the combined features from oximeter and clinical data.METHODS Polysomnography (PSG) data of 680 subjects(age range:18-65 years) were included in this study.The mean pulse oxygen saturation (SpO;),lowest SpO;,oxygen desaturation index(ODI),body mass index(BMI),neck circumference(NC),age,and sex data were selected as parameters for machine learning.XGBoost classifier,support vector machine(SVM),decision tree(DT),random forest(RF) and other common machine learning algorithms were used to determine the optimum algorithm.RESULTS In the 4-classfied multiclass task (non-OSA,mild,moderate and severe OSA),the XGBoost classifier achieved the best performance among all of the algorithms.The accuracy of the proposed XGBoost model was 93.4%.For classification of OSA as normal,mild,moderate or severe using the XGBoost model,the F1-scores were 94.7%,78.7%,86.7%,98.1%,respectively,while the areas under the ROC curve(AUC) were 98.0%,85.0%,92.0%,and 98.0%,respectively.CONCLUSION Therelatively good model results suggest that the XGBoost model can help to reduce a physician’s time required to diagnose OSA compared with having the patient underwent PSG in a clinical setting.
作者
李祖飞
李彦如
施云瀚
韩德民
LI Zufei;LI Yanru;SHI Yunhan;HAN Demin(Department of Otolaryngology Head and Neck Surgery,Beijing Tongren Hospital,Capital Medical University,Key Laboratory of Otolaryngology Head and Neck Surgery(Capital Medical University),Ministry of Education,Beijing,100730,China)
出处
《中国耳鼻咽喉头颈外科》
CSCD
2022年第1期1-5,共5页
Chinese Archives of Otolaryngology-Head and Neck Surgery
基金
国家自然科学基金(81970866)
北京市医院管理局“青苗”计划专项经费(QMS20190202)。
