Precision rehabilitation aims to tailor movement training to improve outcomes. We tested whether proprioceptively-tailored robotic training improves hand function and neural processing in stroke survivors. Using a robotic finger exoskeleton, we tested two proprioceptively-tailored approaches: Propriopixel Training, which uses robot-facilitated, gamified movements to enhance proprioceptive processing, and Virtual Assistance Training, which reduces robotic aid to increase reliance on self-generated feedback. In a randomized controlled trial, forty-six chronic stroke survivors completed nine 2-hour sessions of Standard, Propriopixel or Virtual training. Among participants with proprioceptive deficits, Propriopixel ((Box and Block Test: 7 +/- 4.2, p=0.002) and Virtual Assistance (4.5 +/- 4.4 , p=0.068) yielded greater gains in hand function (Standard: 0.8 +/- 2.3 blocks). Proprioceptive gains correlated with improvements in hand function. Tailored training enhanced neural sensitivity to proprioceptive cues, evidenced by a novel EEG biomarker, the proprioceptive Contingent Negative Variation. These findings support proprioceptively-tailored training as a pathway to precision neurorehabilitation.

翻译：精准康复旨在通过定制化运动训练以改善康复效果。本研究检验了针对本体感觉定制的机器人训练是否能改善脑卒中幸存者的手部功能与神经处理能力。利用机器人手指外骨骼，我们测试了两种本体感觉定制化方法：Propriopixel训练（通过机器人辅助的游戏化运动增强本体感觉处理）和虚拟辅助训练（减少机器人辅助以增强对自我生成反馈的依赖）。在一项随机对照试验中，46名慢性脑卒中幸存者完成了九次、每次2小时的标准训练、Propriopixel训练或虚拟辅助训练。在本体感觉障碍参与者中，Propriopixel训练（方块与积木测试：7 +/- 4.2，p=0.002）和虚拟辅助训练（4.5 +/- 4.4，p=0.068）相较于标准训练（0.8 +/- 2.3个积木）在手部功能改善方面获得更大收益。本体感觉改善与手部功能提升呈正相关。定制化训练通过新型脑电图生物标志物——本体感觉关联负变——证实了其对本体感觉线索神经敏感性的增强作用。这些发现支持将本体感觉定制化训练作为精准神经康复的有效途径。