This article is dedicated to developing a method for diagnosing depression using the analysis of user behavior in a video game on the Unity platform. The method involves employing machine learning to train classification models based on data from gaming sessions of users with confirmed diagnoses of depression. As part of the research, users are engaged in playing a video game, during which their in-game behavior is analyzed using specific depression criteria taken from the DSM-5 diagnostic guidelines. Subsequently, this data is used to train and evaluate machine learning models capable of classifying users based on their in-game behavior. Gaming session data is serialized and stored in the Firebase Realtime Database in text format for further use by the classification model. Classification methods such as decision trees, k-nearest neighbors, support vector machines, and random forest methods have been applied. The diagnostic method in the virtual space demonstrates prospects for remote depression diagnosis using video games. Machine learning models trained based on gaming session data show the ability to effectively distinguish users with and without depression, confirming the potential of this approach for early identification of depressive states. Using video games as a diagnostic tool enables a more accessible and engaging approach to detecting mental disorders, which can increase awareness and aid in combating depression in society.
Keywords: videogame, unity, psychiatric diagnosis, depression, machine learning, classification, behavior analysis, in-game behavior, diagnosis, virtual space