Internet Addiction (IA) is a repetitive and pathological condition characterized by a strong dependence on a specific substance or behavior. IA has become an increasingly prevalent issue, especially among young people and adolescents. To address this issue, a questionnaire was developed according to the criteria of the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) to detect addiction cases among university students in Iraq aged 17–24 years. The results of the questionnaire were analyzed using machine learning (ML) algorithms, specifically (Random Forest (RF), Support Vector Machine (SVM), Neural Network (NN), Linear Discriminant Analysis (LDA), XGBOOST (XGB)). Graph Network Analysis (GNA) was used to analyze and understand patterns and structures in the data, particularly the correlations between questions and answers in the questionnaire, as in the case of this study. The analysis revealed that six questions had strong correlations and could replace all other questions. Based on the utilized data, the SVM algorithm outperformed all other ML algorithms mentioned above in both scenarios (before and after using GNA). When applied to the original questionnaire, an accuracy of 0.973 was achieved, and when applied to the results of the GNA-based questionnaire, an accuracy of 0.991 was obtained. Overall, the classification results of the five ML algorithms improved by an average of 4.4% when using the results of GNA. The SVM algorithm is superior to other algorithms because it works well in cases of binary classification, where it can find the optimal separation between classes.