Blain Stefanie, Mihailidis Alex, Chau Tom
University of Toronto, Biomaterials and Biomedical Engineering, Bloorview Research Institute, 150 Kilgour Road, Toronto, Ontario, Canada.
Med Eng Phys. 2008 May;30(4):498-505. doi: 10.1016/j.medengphy.2007.05.015. Epub 2007 Jul 25.
An embedded multiple-case study was conducted with six able-bodied participants to evaluate the potential of electrodermal activity (EDA) as an alternative access pathway to electronic aids to daily living. Electrodermal signals were recorded while participants alternated between rest and three different mental or breathing exercises. In a subsequent experimental session, the exercise exerting the greatest influence on EDA was used to volitionally generate an 'active' state. Two classification algorithms, namely, a probabilistic classifier and a handcrafted rule base were developed and tailored to each individual's physiological patterns to discriminate between participant states. Through cross-validation, participant state was correctly identified to an accuracy exceeding 80% using either classification algorithm. This result demonstrates that consciously controlled EDA could conceivably serve as a binary switch, and encourages further research towards EDA-based alternative access for people who are locked-in.
对六名身体健全的参与者进行了一项嵌入式多案例研究，以评估皮肤电活动（EDA）作为日常生活电子辅助设备替代接入途径的潜力。在参与者在休息与三种不同的心理或呼吸练习之间交替时记录皮肤电信号。在随后的实验环节中，对EDA影响最大的练习被用于自主产生一种“活跃”状态。开发了两种分类算法，即概率分类器和手工制作的规则库，并针对每个个体的生理模式进行定制，以区分参与者的状态。通过交叉验证，使用任何一种分类算法对参与者状态的正确识别准确率均超过80%。这一结果表明，有意识控制的EDA可以想象地用作一个二元开关，并鼓励针对闭锁人群基于EDA的替代接入进行进一步研究。
