Identifying EEG-based neurobehavioral risk markers of gaming addiction using machine learning and iowa gambling task

Comprehensive Summary

This study by Kornev et al. examines EEG-based biomarkers in healthy participants to identify early risk markers for gaming addiction by classifying them into groups based on behavioral patterns observed during the Iowa Gambling Task (IGT). In the study, EEG and IGT data were collected simultaneously from healthy participants, with the IGT serving as a cognitive challenge to force decision-making under uncertainty and risk. The EEG signals were segmented into event-related potentials (ERPs), preprocessed, and temporal features were extracted for analysis. Participants were then placed into groups based on their IGT performance using a 10-point score threshold, using IGT performance as an intermediate behavioral phenotype to identify addiction-prone profiles. The researchers were able to successfully use machine learning algorithms to classify participants into two groups based on their behavioral patterns during the IGT, with one group exhibiting tendencies consistent with addiction-prone profiles. They also found that EEG-based features extracted from the event-related potentials during decision-making under uncertainty were able to distinguish between these groups. The machine learning approach enabled detection of markers of maladaptive behavior in a non-clinical sample, showing that EEG biomarkers could identify early risk patterns of gaming disorder. The study demonstrates that the IGT can be used not to diagnose gaming disorder directly, but rather to group participants who are exhibiting behavioral tendencies consistent with addiction-prone profiles, allowing for more sensitive machine learning detection prior to clinical symptoms showing.

Outcomes and Implications

Internet gaming disorder, gaming disorder, and internet addiction represent behavioral patterns with significant psychological and neurological consequences, with affected individuals often disengaging from routine life in order to game and experiencing distress when interrupted, impacting family life and overall wellbeing. Timely and objective detection methods are essential for early intervention, as current diagnostic approaches often rely on self-reported symptoms and clinical presentation after significant impairment has already developed. This research provides an objective, neurophysiology-based approach to identifying individuals at risk. The use of EEG combined with machine learning offers a quantifiable biomarker approach that could supplement traditional behavioral assessments in clinical settings. The methodology demonstrates potential for screening and monitoring vulnerability to gaming disorder. However, the study's focus on solely healthy participants suggests that further validation in clinical populations and longitudinal studies would be necessary before widespread clinical adoption.