Smartphones dependency risk analysis using machine-learning predictive models.

Area of Science:

• Digital Health
• Psychology
• Computer Science

Background:

• Extensive technology use, particularly smartphones, contributes to sedentary lifestyles and new behavioral addictions.
• Smartphone dependency is a growing concern, disproportionately affecting young populations and potentially leading to adverse mental and physical health outcomes.
• Current assessment of smartphone dependency relies on subjective self-reports and expert opinions, necessitating more objective diagnostic tools.

Purpose of the Study:

• To propose and evaluate a data-driven prediction model for smartphone dependency using machine learning techniques.
• To assess the accuracy of various machine learning classifiers in predicting smartphone dependency.
• To explore the utility of self-reported information in identifying individuals with smartphone dependency.

Main Methods:

• An analytical retrospective case-control study was conducted with 1228 university students in Cali, Colombia.
• Machine learning classification methods, including Random Forest, Logistic Regression, and Support Vector Machines, were applied.
• Stratified k-fold cross-validation was employed to estimate prediction accuracy for smartphone dependency, musculoskeletal symptoms, and risk factors.

Main Results:

• Random Forest, Logistic Regression, and Support Vector Machine classifiers demonstrated the highest prediction accuracy, ranging from 76% to 77%, for smartphone dependency.
• The study confirmed that self-reported information can provide valuable insights for accurately predicting smartphone dependency.
• The findings underscore the effectiveness of machine learning models in identifying potential cases of smartphone addiction.

Conclusions:

• Self-reported data, when analyzed with machine learning, can effectively predict smartphone dependency.
• The developed prediction model offers a promising approach for early identification and intervention of smartphone addiction.
• Future research should incorporate objective measures to further enhance prediction accuracy and mitigate the negative impacts of smartphone dependency.