Game-Based CBT with Artificial Intelligence: A Method for Correcting Cognitive Distortions by Revisiting the Past (Preprint)

Abstract

UNSTRUCTURED Given that the most used therapies in psychiatry, Cognitive Behavioral Therapy (CBT) and Mindfulness-Based Cognitive Therapy (MBCT), were developed to treat depression, a new approach is needed for disorders such as Internet Gaming Disorder and Gambling Disorder, where the issue lies not in negative emotions themselves but in their expression through problematic behaviors. These behaviors act as the patient’s form of “self-prescription” to relieve stress, making it essential to propose alternative coping strategies that can effectively alleviate negative emotions. This paper proposes a novel game-based digital therapy method that reflects each patient’s individual “automatic thoughts,” based on an emotional theory expanded into a four-dimensional cognitive concept that incorporates both the feedback from coping strategies and the subjective interpretation of that feedback. An LLM agent based on a large language model and a doctor agent based on rule-based machine learning work complementarily, enabling the patient to safely and repeatedly practice active coping strategies in crisis situations.