| 摘 要: 睡眠呼吸暂停综合征(SleepApnedSyndrome,SAS)对生活质量影响显著。多导睡眠图是检测该病最有效的方法,但因其体积大限制了家庭使用。提出一种适用于家庭环境的自动化检测方法,利用微弯曲光纤传感器采集睡眠中的生物振动数据,获得心冲击描记图(Ballistocardiogram,BCG)信号,采用小波散射变换处理BCG信号,提取特征并输入随机森林进行心电图信号片段分类。通过10倍交叉验证和软投票法验证,准确率分别达到89.55%和88.56%。结果表明,小波散射变换提取的高阶特征能够有效分类睡眠呼吸暂停片段。 |
| 关键词: 睡眠呼吸暂停综合征 心冲击描记图 小波散射变换 特征提取 随机森林 |
|
中图分类号: TP701
文献标识码: A
|
| 基金项目: 国家自然科学基金项目(32360437);甘肃省高等学校创新基金项目(2021A-056) |
|
| Detection of Sleep Apnea Syndrome Based on Random Forest and Wavelet Scattering Transform |
|
LI Xuanyi1, ZHU Yaodong1,2, ZHAO Weilan1, FAN Ziyan1
|
(1.School of Information Science and Engineering, Zhejiang Sc-i Tech University, Hangzhou 310018, China;
2. College of Information Science and Engineering, Jiaxing University, Jiaxing 314001, China)
575697837@qq.com; zhuyaodong@163.com; 2484497912@qq.com; 202230603058@mails.zstu.edu.cn
|
| Abstract: Sleep Apnea Syndrome (SAS) significantly impacts quality of life. Polysomnography is the most effective clinical diagnostic method but is unsuitable for home use due to its bulkiness. This study proposes an automated detection approach for home environments, utilizing a microbend fiber optic sensor to acquire biological vibration signals during sleep and generate Ballistocardiogram (BCG) data. Wavelet scattering transform was applied to process BCG signals, extracting features subsequently input into a random forest classifier for electrocardiogram (ECG) segment classification.Through 10-fold cross-validation and soft voting methods, accuracies reached 89. 55% and 88.56% , respectively. The results demonstrate that highe-r order features extracted by wavelet scattering transform can effectively classify sleep apnea segments. |
| Keywords: sleep apnea syndrome ballistocardiogram wavelet scattering transform feature extraction random forest |
