This research aims to study the psycho-social impact of video games on K12 students. For this, a probabilistic topic modelling analysis method based on text mining approach has been performed. This process is based on nodes' connectivity and it has been developed through K Means approach; by launching the Jenks-Breaks algorithm. The sex differences are calculated according to a nonlinear dynamics approach based on Hurst exponent and multifractal function and the influence of time with the application of the Sobel test. The results show which are the most used video games by K12 and their psycho-social impact on students based on four categories: (1) boredom and sadness, (2) happiness and socialization, (3) immersion, and (4) families' conflicts. There are significant differences between boys and girls depending on the games they use, a factor that increases when playing more than two hours a day. For boys, games like FIFA and Fortnite produce higher levels of immersion and family conflict, while for girls, games are perceived as sources of greater happiness and a means to reinforce friendship and camaraderie, particularly with games like Brawl Stars, Rocket League, and PKXD.