THE EFFECTS OF AI-DRIVEN SERIOUS VIDEO GAMES ON FACIAL EXPRESSION ABILITIES AND ACADEMIC PERFORMANCE OF CHILDREN WITH AUTISM SPECTRUM DISORDER: AN EMPIRICAL STUDY

Abstract

Background: Children with autism spectrum disorder (ASD) often face challenges in creating facial expressions that appropriately respond to social cues, which can affect both their social interactions and academic performance. While many serious games have been designed to assist players with facial expression identification, fewer have focused on training children to generate facial expressions suited to specific social settings. This study explores an AI-driven, gamified approach aimed at enhancing both facial expression abilities and academic outcomes for children with ASD.

Method: A video game was developed featuring a 3D virtual world that presents social scenarios in which the player must express emotions such as happiness, anger, or sadness. The game includes real-time feedback through an integrated AI-powered facial expression detection system, which allows players to assess and improve their facial expressions. The AI-driven system utilizes advanced facial recognition technology to analyze and provide immediate, individualized feedback to the player, enhancing the learning experience. Special attention was given to the aesthetic and motivational elements of the game's design, incorporating gamified features such as levels, rewards, and challenges to maintain player engagement. A brief feasibility study involving autistic children (N = 25) was conducted to evaluate their perceptions of the game, the effectiveness of real-time AI feedback, and the impact of the game on academic performance.

Results: Participants showed significant improvements in producing facial expressions in response to the AI algorithm's real-time feedback during the training sessions. This suggests that children were able to comprehend the complexity of creating appropriate facial expressions and that the feedback provided by the AI algorithm was effective in aiding this learning process. The AI-driven nature of the game provided tailored feedback that adapted to each player's performance, allowing for more targeted learning. Participants also reported positive subjective experiences regarding the game's playability, motivation, aesthetic appeal, and ergonomics, emphasizing the effectiveness of the gamified elements such as progression through levels and the achievement of rewards. Moreover, results revealed that 60% of the participants demonstrated an increase in their academic performance after engaging with the serious game, indicating a potential link between improved emotional intelligence and academic success.

Conclusion: The beta version of the game appears to hold promising potential, and further research is recommended to develop additional games and scenarios to enhance the learning experience. Extending gameplay duration could allow for comprehensive clinical validation, thereby better supporting children with ASD in both social and academic aspects of their lives. Future studies should focus on refining the AI algorithms and expanding the gamified elements to create a more robust, immersive, and clinically validated intervention for children with ASD.