基于多特征信息的双层睡眠分期模型OA北大核心CSTPCD
Dual-layer sleep stage classification model based on multiple features
针对睡眠相关疾病的诊断和治疗过程中睡眠分期的问题,使用ADASYN算法对睡眠样本进行类平衡,有效利用复合多尺度排列熵、样本熵和不同频段的能量值等一系列能够反映不同睡眠阶段信息的特征进行训练.为了提高分类中易混淆睡眠阶段的分类识别性能,构建了一种双层分类模型,使睡眠分期的五分类问题转化为两个三分类问题.将所提出的方法在Sleep-EDF数据集上进行验证,结果表明:所提出的模型对健康受试者的准确率可达到88.3%,较以往模型提高1%～3%,其中N1阶段的分类准确率可达69.5%,较以往模型提高约10%,证明本双层分类模型优于传统的睡眠分类模型.
Aiming at the issue of sleep staging in diagnosis and treatment of sleep-related disorders,the ADASYN algorithm was used to balance the sleep samples,and a series of features reflecting different sleep stages,such as composite multiscale permutation entropy,sample entropy,and energy values in different frequency bands was effectively utilized for training.To enhance the recognition performance of easily confused sleep stages in classification,a dual-layer classification model was constructed,transforming the five-class sleep staging problem into two three-class problems.The proposed method was validated on the Sleep-EDF dataset.Results show that the proposed model achieves an accuracy of 88.3%for healthy subjects,which is an improvement of 1%～3%compared to previous models,with a classification accuracy of N1 stage reaching 69.5%,which is an increase of about 10%compared to previous models,so it is proved that the proposed dual-layer classification model outperforms traditional sleep classification models.
张虹淼;王梓;廖肖剑;李伟达
苏州大学机电工程学院江苏省先进机器人技术重点实验室,江苏 苏州 215021
基础医学
睡眠分期脑电信号特征提取复合多尺度排列熵双层分类模型
sleep stage classificationelectroencephalogramfeature extractioncomposite multiscale permutation entropydual-layer classification model
《华中科技大学学报（自然科学版）》 2024 (005)
64-69 / 6
广东省重点领域研发资助项目(2019B090915002).
评论