本文采用脉搏波信号研究了警觉度的变化规律。本研究共招募 10 名受试者参加了持续 95 min 的警觉度“时钟测试”（MCT）。试验过程中，受试者们通过卡罗林斯卡嗜睡量表（KSS）和斯坦福嗜睡量表（SSS）主观评价了清醒程度，同时记录了所有受试者的目标反应时间、目标识别正确率和脉搏波信号。结果表明，根据主观量表得分和受试者的行为学数据可以将警觉度定标为 3 个水平：前 30 min 为高警觉度水平；中间 30 min 为一般警觉度水平，后 30 min 为低警觉度水平。此外，脉搏波信号的时域特征，如：次级波峰幅值、波峰幅值、次级波峰潜伏期，随警觉度的降低而减小，而波谷幅值随警觉度的降低而增大；频域特征：8.600～9.375 Hz、11.720～12.500 Hz、38.280～39.060 Hz 和 39.060～39.840 Hz 这 4 个子频带的能量概率也随警觉度的降低而减小。最后，在上述 8 个特征建立的模型中，10 名受试者三分类正确率的平均值高达 88.7%。本文的研究结果证实了脉搏波在警觉度评估上的可行性，为警觉度的实时监测提供了新的思路。
This paper studied the rule for the change of vigilance based on pulse wave. 10 participants were recruited in a 95-minute Mackworth clock test (MCT) experiment. During the experiment, the vigilance of all participants were evaluated by Karolinska sleepiness scale (KSS) and Stanford sleepiness scale (SSS), and behavior data (the reaction time and the accuracy of target) and pulse wave signal of the participants were recorded simultaneously. The result indicated that vigilance of the participants can be divided into 3 classes: the first 30 minutes for high vigilance level, the middle 30 minutes for general vigilance level, and the last 30 minutes for low vigilance level. Besides, time domain features such as amplitude of secondary peak, amplitude of peak and the latency of secondary peak decreased with the decrease of vigilance, while the amplitude of troughs increased. In terms of frequency domain features, the energy of 4 frequency band including 8.600 ~ 9.375 Hz, 11.720 ~ 12.500 Hz, 38.280 ~ 39.060 Hz and 39.060 ~ 39.840 Hz decreased with the decrease of vigilance. Finally, under the recognition model established by the 8 characteristics mentioned above, the average accuracy of three-classification results over the 10 participants was as high as 88.7%. The results of this study confirmed the feasibility of pulse wave in the evaluation of vigilance, and provided a new way for the real-time monitoring of vigilance.
引用本文： 曹勇, 焦学军, 潘津津, 姜劲, 傅嘉豪, 徐凤刚, 杨涵钧. 基于脉搏波的警觉度检测研究. 生物医学工程学杂志, 2017, 34(6): 817-823. doi: 10.7507/1001-5515.201704071 复制