IEEE Trans Neural Syst Rehabil Eng. 2018 Feb;26(2):487-497. doi: 10.1109/TNSRE.2017.2728140. Epub 2017 Jul 18.
This paper illustrates a new architecture for a human-humanoid interaction based on EEG-brain computer interface (EEG-BCI) for patients affected by locked-in syndrome caused by Amyotrophic Lateral Sclerosis (ALS). The proposed architecture is able to recognise users' mental state accordingly to the biofeedback factor , based on users' attention, intention, and focus, that is used to elicit a robot to perform customised behaviours. Experiments have been conducted with a population of eight subjects: four ALS patients in a near locked-in status with normal ocular movement and four healthy control subjects enrolled for age, education, and computer expertise. The results showed as three ALS patients have completed the task with 96.67% success; the healthy controls with 100% success; the fourth ALS has been excluded from the results for his low general attention during the task; the analysis of factor highlights as ALS subjects have shown stronger (81.20%) than healthy controls (76.77%). Finally, a post-hoc analysis is provided to show how robotic feedback helps in maintaining focus on expected task. These preliminary data suggest that ALS patients could successfully control a humanoid robot through a BCI architecture, potentially enabling them to conduct some everyday tasks and extend their presence in the environment.
本文提出了一种基于脑电-脑机接口（EEG-BCI）的人机交互新架构，用于治疗肌萎缩性侧索硬化症（ALS）导致的闭锁综合征患者。所提出的架构能够根据生物反馈因素识别用户的精神状态，基于用户的注意力、意图和焦点，引发机器人执行定制行为。实验在 8 名受试者中进行：4 名处于接近闭锁状态且眼部运动正常的 ALS 患者和 4 名年龄、教育程度和计算机专业知识相匹配的健康对照组。结果表明，3 名 ALS 患者以 96.67%的成功率完成了任务；健康对照组的成功率为 100%；第 4 名 ALS 患者由于在任务过程中注意力普遍较低而被排除在结果之外；因子分析表明，ALS 患者的注意力明显强于健康对照组（81.20%比 76.77%）。最后，提供了一个事后分析来展示机器人反馈如何帮助维持对预期任务的关注。这些初步数据表明，ALS 患者可以通过 BCI 架构成功地控制人形机器人，从而使他们能够完成一些日常任务，并在环境中扩展他们的存在。
