Markov Chain Modeling of Facial Emotions’ Dynamics in Video Games

Serious Games (SG) are digital applications that can be used in healthcare to educate individuals on emotional and cognitive skills. This study proposes an innovative method to characterize and understand facial emotion (FE) in healthy subjects while playing three consecutive trials of two video games, namely the Snake (S) game and the Matching Pairs (MP) game. WAFER-ET, a web-based real-time platform integrating an AI-based emotion model, was used to record and classify individuals’ FE and a discrete state-space Markov Chain (MC) process was used to represent the dynamic evolution of FE in terms of transition matrices (TM) for both games. TM-related features were statistically compared in terms of individual performance. Results report higher entropy rates and lower mean hitting times for the happy FE values in well-performing individuals at the S game. In contrast, lower spectral gap values are observed in well-performing players at the MP game. Though preliminary, this study hints at the dynamics of FE as being relevant for cognitive and emotional skills evaluation in individuals.