Dynamic Difficulty Adjustment of Rehabilitation Games using Reinforcement Learning

Research has shown that the design of a rehabilitation task as a therapeutic game can result in a motivating rehabilitation environment. Generally, the difficulty level of a therapeutic game is regulated manually by a therapist. However, home-based rehabilitation games require a technique for automatic difficulty adjustment. This paper proposes a personalized difficulty adjustment technique for a rehabilitation game that automatically regulates difficulty settings based on a patient’s skills in real-time. To this end, ideas from reinforcement learning are used to dynamically adjust the difficulty of a game. We show that difficulty adjustment is a multiple-objective problem, in which some objectives might be evaluated at different periods. To address this problem, we propose and use Multiple-Periodic Reinforcement Learning that makes it possible to evaluate different objectives of difficulty adjustment in separate periods. The results of experiments show that this technique outperforms traditional Multiple-Objective Reinforcement Learning in terms of user satisfaction parameters.