针对个体差异的心理压力评估

发布时间：2018-03-02 06:17

本文关键词： 肌电心理压力评估个体差异支持向量机权重　出处：《中国生物医学工程学报》2014年01期 　论文类型：期刊论文

【摘要】：慢性心理压力会带来一系列的病理、生理风险,直接影响健康。有效地评估心理压力,一直是心理压力研究中的热点问题。在心理压力评估过程中,个体差异是影响评估效果的关键。本研究针对评估心理压力/非压力反应中个体差异问题,以表面肌电信号作为评估参数,以高校即将毕业的学生人群为对象,提出了一种改进的支持向量机心理压力评估算法。算法通过对样本聚类,并将聚类信息赋予支持向量机的损失函数,实现训练样本的筛选,针对筛选后出现两类样本不平衡问题,为损失函数赋予权重来降低分类器的预测倾向性,减少训练模型的误差,补偿不平衡样本数据所造成的影响。心理压力评估分类正确率由改进前的70.34%,提高到79.31%,算法运行时间由改进前的2026.5 s减少到541.3 s。结果表明,该算法可以有效地解决个体差异对于心理压力评估效果的影响,同时降低了分类器的计算复杂度,为心理压力评估中个体差异研究提供一种可行的方案。
[Abstract]:Chronic stress can bring a series of pathological, physiological risks and directly affect health. Effective assessment of psychological stress has always been a hot issue in the study of psychological stress. The individual difference is the key to the evaluation effect. In order to evaluate the individual difference in psychological stress / non-stress response, the surface electromyography (EMG) signal is taken as the evaluation parameter, and the students graduating from the university are taken as the objects. An improved support vector machine (SVM) psychological stress assessment algorithm is proposed. By clustering the samples and giving the clustering information to the loss function of the SVM, the training samples can be screened. Aiming at the problem of unbalance of two kinds of samples after screening, the loss function is given weight to reduce the forecast tendency of classifier and the error of training model. The accuracy rate of psychological stress assessment was increased from 70.34 to 79.31, and the running time of the algorithm was reduced from 2026.5 s to 541.3 s. The algorithm can effectively solve the impact of individual differences on the psychological stress assessment, and reduce the computational complexity of the classifier. It provides a feasible scheme for the study of individual differences in psychological stress assessment.
【作者单位】：燕山大学生物医学工程研究所;河北省测试计量技术及仪器重点实验室;北京工业大学生命科学与生物工程学院;
【基金】：河北省优秀专家出国资助项目河北省教育厅重点项目(ZD2010115)
【分类号】：R318

【相似文献】