Event-Driven EEG Analysis and Machine Learning for Understanding Gaming Disorder in Actual Game Play

초록

Although gaming offers diverse benefits, its widespread use has also raised concerns regarding atypical patterns, such as gaming disorder (GD), highlighting the need for objective assessment methods. This study examines neurophysiological differences between individuals with GD and healthy controls (HCs) during gameplay. Electroencephalogram (EEG) signals are analyzed after kill and death events in the game League of Legends, across four post-event windows (2, 5, 7, and 10 s). Several machine learning models are applied to classify the GD and HC groups and identify the most relevant windows and EEG markers for discriminating between these groups. The random forest model achieved 88.4% accuracy and 95.0% area under the curve in the 10-s window. The feature importance analysis highlighted theta activity in the O2 channel and beta and high-beta activity in the F3 channel, although the O2-theta effect was not statistically significant. This discrepancy demonstrates how EEG responses combined with machine learning can uncover latent patterns after in-game events in player-game interactions and behavioral patterns.

키워드