Emotion-Adaptive Somatosensory Gaming for Hand Rehabilitation: Interactive Design and EEG-based Assessment

Conventional hand rehabilitation with mild stroke often suffers from mechanical monotony and a lack of motivational appeal, leading to engagement and a critical neglect of affective dimensions in therapy. To overcome efficacy limitations rooted in low adherence, this study establishes a closed-loop affective adaptation framework grounded in Flow Theory. By integrating a game-based Dynamic Modulation of Challenge (DMC) mechanism with electroencephalography (EEG)-driven affective computing, we developed Rhythm Orchard, a serious game that dynamically synchronizes task difficulty with real-time emotional feedback. Through iterative user-centered research, twelve evidence-informed design principles were formulated. Further more, an EEG-based emotion recognition framework, utilizing a lightweight multi-dimensional attention network (LMDA-Net) and differential entropy features, was implemented to ensure robust, low-latency estimation of affective states. Controlled experimental results demonstrate that Rhythm Orchard significantly enhances participants’ emotional valence, physiological arousal, and subjective immersion, thereby fostering higher therapeutic engagement and improved functional outcomes. These findings underscore the transformative potential of intelligent games in geriatric stroke rehabilitation.