Construction of diagnostic model of depression in insomnia patients based on polysomnography data
曹宁 1张慧如 2牛丽薇 1赵瑞2
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. 内蒙古医科大学公共卫生学院,呼和浩特 010110
2. 内蒙古自治区精神卫生中心
折叠
摘要
目的 通过机器学习算法对失眠患者的多导睡眠监测(polysomnography,PSG)数据进行挖掘,建立失眠患者抑郁症的诊断模型,为失眠患者的抑郁症诊断提供科学依据.方法 选择2023年1～12月在内蒙古自治区精神卫生中心进行PSG的失眠住院与门诊患者共2162例,抑郁症根据《国际疾病与相关健康问题统计分类第10版》(International Statistical Classification of Diseases and Related Health Problems,10th version,ICD-10)进行诊断.收集患者的一般情况与PSG资料,分别基于logistic回归、支持向量机、随机森林、自适应提升、极限提升树、朴素贝叶斯等6种算法构建失眠患者抑郁症的诊断模型.结果 纳入的失眠患者中,40.1％(868例)的患者合并抑郁症.6种模型中,logistic回归和随机森林模型的受试者操作特征曲线(receiver operating characteristic curve,ROC curve)的曲线下面积(area under the curve,AUC)值最高,分别为0.825和0.823,综合分类性能更优.结论 Logistic回归和随机森林模型对失眠患者中的抑郁症人群有良好的诊断效能.
Abstract
Objective To establish a diagnostic model for depression in insomnia patients by mining polysomnography (PSG) data of insomnia patients with machine learning algorithms,and to provide a scientific basis for the diagnosis of depression in insomnia patients. Methods According to the inclusion and exclusion criteria,2162 insomnia inpatients and outpatients who attended the Inner Mongolia Autonomous Region Mental Health Center from January to December 2023 and underwent polysomnographic monitoring were included,and depression was diagnosed using the International Statistical Classification of Diseases and Related Health Problems,10th version (ICD-10). The general condition and PSG data of the patients were collected. Six algorithms—logistic regression (LR),Support vector machines (SVM),Random forest (RF),Adaptive Boosting (AdaBoost),Extreme Gradient Boosting (XGBoost) and Naive Bayes (NB)—were used to build the diagnostic model of depression in insomnia patients after the patients' general condition and PSG data were gathered. Results Among the enrolled patients with insomnia,40.1％ had comorbid depression. Among the six models,LR and RF exhibited the highest values of area under the curve (AUC) of receiver operating characteristic (ROC),at 0.825 and 0.823,respectively,indicating superior overall classification performance. Conclusion Logistic regression and random forest modeling have good diagnostic efficacy in the population of insomniacs with depression.
关键词
失眠/抑郁症/多导睡眠监测/机器学习/诊断模型/逻辑回归/随机森林模型
Key words
Insomnia/Depression/Polysomnography/Machine learning/Diagnostic model/Logistic regression/Random forest model
维普
万方数据