基于穿戴式生理信号监测的学生课堂学习状态评测研究

Objective and accurate assessment of the students’ attention, emotion and relaxation in normal classroom could provide invaluable scientific evidence for teaching quality assessment and class teaching improvement. Continuous monitoring of physiological signals using wearable devices is a helpful approach to assess and follow up students’ learning state dynamically, and the accurate representation of learning state based on physiological signal feature detection is the key point. In this project, we propose to collect and analyze real-time EEG and ECG signals from students in normal classroom within a whole semester to study the association of individual physiological systems’ functional state and learning state, and the association of unique information pattern of individual physiological signal and specific learning state, to obtain potential ‘biomarkers’ of different state of students' engagement. We aim to develop novel and accurate algorithms to quantify in-class student engagement through studying students’ attention and emotions both short term and long term from physiological signals. We plan to integrate signal analysis methods and deep learning through ‘wide & deep’ learning models to capture the physiological and psychological dynamics on multiple time-scale. Through longitudinal follow up and real time monitoring of physiological states of students and translation of such information to in-class engagement, we aim to provide scientific evidence for teaching quality assessment and education quality improvement.

真实课堂情景下学生学习专注度、情绪、放松度等学习状态的客观、精准评测关系到教育质量监测和课堂教学改革。穿戴式生理信号监测是量化评测并动态跟踪学生课堂学习状态的重要途径，基于生理信号特征检测的个体学习状态精准表征是其中关键环节。本项目拟通过可穿戴技术，在一个学期内动态采集学生在真实课堂学习过程中的实时脑电和心电信号，通过研究个体生理系统的功能状态与学习状态之间的映射关系，和个体生理信号的特种变化模式与特定学习状态之间的对应关系，提出与特定学习状态对应的新的特异性生理指标，将课堂学习过程中学生的生理动态特性映射到学习状态。并进一步基于深度学习的Wide&Deep学习模型，分别建立基于短时和长时数据分析的当前和阶段学习状态评测模型，从多个时间尺度精准刻画学习状态的动态变化规律，实现对课堂学习状态的历史动态准确跟踪，为教育质量监测和教学评估提供可靠的科学依据。

本项目旨在真实课堂情景下利用穿戴式生理信号监测技术对学生的专注度、情绪、放松度/压力等学习状态进行客观、精准评测。项目执行过程中重点研究并实现了1）基于可穿戴生理信号监测的团体版学生课堂学习状态评测软硬件系统，建立了适用于普通教室中30人及以上课堂的学生脑电、心电、脉搏波数据同步采集和传输平台；2）基于穿戴式生理信号（脑电、心电、脉搏波）的学生学习状态评测（专注度、情绪、压力/放松度、理解度等），将学习过程中学生的生理动态特性映射到学习状态，分别提出了与专注度、情绪、放松度/压力等学习状态对应的新的特异性生理指标；3）真实课堂情景下穿戴式生理数据的准确获取与数据库建设及标注，并完成了可穿戴生理信号的自适应滤波和信号质量评估相关算法。项目执行期间发表了SCI论文4篇，EI论文2篇，其他期刊和会议论文6篇；申请发明专利3项；培养博士研究生2名，硕士研究生5名。本项目积累的数据库和研究结果将为教育质量监测和教学改革提供评测工具和科学依据。