Entry-Level Game Difficulty Estimation Based on Visuoperceptual Profiles of Children with Cerebral Visual Impairment

Cerebral visual impairment (CVI), a leading cause of visual impairment in developed countries, has no standard treatment and many different manifestations. Long-term individualized therapy is crucial for children with CVI to develop coping strategies and improve functional limitations. Serious games are becoming a part of many different types of therapy and rehabilitation, including physical rehabilitation, cognitive behavioral therapy, vision therapy, etc. However, the design of the serious games used in the context of rehabilitation and therapy must ensure that the games fulfill the therapeutic goals and also adapt to the skills and abilities of the players. Entry-level estimation, i.e., placing the player in a correct initial game level the first time the player engages with the game, is an important aspect of individualization and personalization of serious games.

In this paper, we discussed the importance of adaptivity and personalization of serious games in children with CVI, specifically focusing on entry-level difficulty estimation and the comparison of two entry-level estimation models. The first model was defined based on discussions with experts in the CVI therapy field and the second was a linear regression model trained on in-game collected data. We evaluated the two models on several performance metrics and compared how the entry-level estimation using the first versus second formula would differ for the children participating in our currently ongoing Randomized Control Trial.