EUVIMEDEUVIMED entdecken
Crossref

Machine Learning-Based Prediction of Smartphone Addiction Using Behavioral Usage Features

Hasanain Hazim Azeez

Journal of Al-Farabi for Engineering Sciences · 12.05.2026 · 5 (1) · 94-107

DOI: 10.59746/ay1jrx88 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Al-Farabi for Engineering Sciences, online erschienen am 12.05.2026. Verfasst von Hasanain Hazim Azeez.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The rapid growth of smartphone usage has raised serious concerns regarding problematic and addictive usage behaviors, which may negatively affect sleep quality, productivity, and psychological well-being. Despite increasing research attention, existing studies often rely on self-reported scales or limited analytical approaches, reducing objectivity and generalizability. This study aims to develop a data-driven framework for predicting smartphone addiction using behavioral usage features and to identify the most influential predictors of addictive behavior. A publicly available dataset consisting of 500 users (Kaggle, 2024) was analyzed, encompassing demographic and behavioral variables including age, daily screen time, social media notification frequency, sleep duration, and number of installed applications. Four supervised machine learning classifiers—Random Forest (RF), Support Vector Machine (SVM), Logistic Regression (LR), and Decision Tree (DT)—were trained and evaluated under identical conditions using an 80/20 stratified split with 10-fold cross-validation. Class imbalance was addressed using SMOTE applied exclusively to the training set. Results indicate that daily screen time, notification frequency, and sleep duration are the strongest predictors of smartphone addiction. The Random Forest model achieved the highest performance (Accuracy = 88.4%, F1-score = 0.898, AUC = 0.923). Correlation analysis confirmed strong associations between addiction status and screen time (r = 0.721, p < 0.01), notification frequency (r = 0.653, p < 0.01), and a significant negative association with sleep duration (r = −0.611, p < 0.01). Approximately 58.2% of users in the sample were classified as addicted. The findings demonstrate that behavioral usage data can effectively support automated prediction of smartphone addiction and contribute to the development of real-time digital health monitoring tools. This study provides a reproducible comparative machine learning framework and highlights key behavioral indicators suitable for early detection and clinical screening applications.

Zitieren

APA

Azeez, H. H. (2026). Machine Learning-Based Prediction of Smartphone Addiction Using Behavioral Usage Features. Journal of Al-Farabi for Engineering Sciences, 5(1), 94-107. https://doi.org/10.59746/ay1jrx88

Vancouver

Azeez HH. Machine Learning-Based Prediction of Smartphone Addiction Using Behavioral Usage Features. Journal of Al-Farabi for Engineering Sciences. 2026;5(1):94-107. doi:10.59746/ay1jrx88

BibTeX
@article{Azeez2026,
  author = {Azeez, Hasanain Hazim},
  title = {Machine Learning-Based Prediction of Smartphone Addiction Using Behavioral Usage Features},
  journal = {Journal of Al-Farabi for Engineering Sciences},
  year = {2026},
  volume = {5},
  number = {1},
  pages = {94--107},
  doi = {10.59746/ay1jrx88}
}
Kostenlos starten →