基于模糊支持向量机睡眠呼吸暂停综合征分类.pdfVIP

基于模糊支持向量机的睡眠呼吸暂停综合征的分类 刘素娟 江丽仪 吴效明 (华南理工大学生物医学工程系，广州 510006) 摘要：采用去趋势波动分析法(DFA)和自回归模型谱分析法研究心率变异性与睡眠时相的相关性，计算了38 名健康者和28 名不同程度睡眠呼吸暂停综合征(SAS)在各个睡眠阶段的标度指数及心率变异的频域特征；并比较了健康与 SAS 的各参数 间的差异性，提出新的 SAS 初筛的特征参数。应用模糊支持向量机(FSVM)将正常数据与异常数据分类，实现对 SAS 的自 动初步筛选。通过实验测试，其分类正确率达到 93.94% 。因此可协助人工诊断实现社区大规模的 SAS 初步筛选，减少医 务工作者的工作量，提高诊断效率。 关键字：心率变异性 去趋势波动分析 睡眠时相 模糊支持向量机 睡眠呼吸综合征 Based on Fuzzy Support Vector Machine to classify Sleep Apnea Syndrome Sujuan-Liu Liyi-Jiang Xiaoming-Wu (Department of Biomedical Engineering, South China University of Technology, Guangzhou 510006, China) Abstract ：This paper studies the correlation between heart rate variability and sleep phase using detrended fluctuation analysis(DFA) and AR model spectral analysis methods, and calculates the fractal exponents in different sleep stage and frequency domain characteristics of heart rate variability, which belong to the 38 healthy subjects and 28 SAS patients. It also compares the parameter difference between the health and sleep apnea syndrome(SAS); proposing new preliminary screening features of SAS. Based on Fuzzy Support Vector Machine(FSVM), normal and abnormal data are classified to achieve automatic preliminary screening for SAS. From the experiment, the classification accuracy rate can reach 93.94%. It demonstrates that what this paper studies can assist the artificial diagnosis to screen SAS in large-scale community and improve the diagnostic efficiency. Key Words: heart rate variability, detrended fluctuation analysis, sleep phase, fuzzy support vector machine, sleep apnea syndrome 1 引言 睡眠呼吸暂停综合征（Sleep Apnea Syndrome，SAS ）是指在睡眠时的某一段时间内没有吸气和呼气 气流，并由此引发起血氧下降的一种疾病。