A Difficulty Model Based on Movement Abilities for a Post-Stroke Rehabilitation Game (Preprint)

BACKGROUND Therapeutic games and Virtual Reality are popular tools for stroke rehabilitation. However, to be effective, the exercises must be repetitive, intensive, task-oriented and should present challenges adapted to the patient’s abilities and condition. OBJECTIVE In this article, we propose a model to control the difficulty in post-stroke rehabilitation games, based on movement abilities and endurance. We apply this in a forearm pronation-supination game. METHODS A preliminary experiment with 26 chronic stroke patients and 32 healthy participants studies the usability of the system and the impact of our difficulty model on performance, as well as the acceptance and motivation of users, and the perception of the difficulty and efforts. RESULTS Our system has excellent usability according to the System Usability Scale (SUS) scores. Objective and subjective results indicate that our difficulty model is coherent and has an impact on the performance of participants, though in limited proportions. Patients and healthy subjects have a strong and identical interest which suggests that therapeutic gaming is well designed to motivate users. CONCLUSIONS We introduce a difficulty model based on motor abilities that can be used by various therapeutic games. This formal description allows to include a part of the therapist’s expertise through coefficients that can be adjusted according to the therapeutic objectives and patient profiles. The next steps are to make a framework from the model and to pair it with automatic difficulty adjustment. CLINICALTRIAL CER-Paris-Saclay-2018-040