Machine Learning Approach to Analyse the Mental Health and Predict Nomophobia

With the growing concern in today's digital world, the fear of not having a mobile phone, which is termed as nomophobia, has become an alarming issue, especially for the youth, because of its drastic effect on mental health in terms of anxiety, stress, and depression. This research develops a predictive framework to identify severe nomophobics by utilizing advanced machine learning techniques. A sample of 841 students from university in Kosovo completed the survey through purposive and stratified sampling to ensure a balanced dataset. Due to the robustness of handling non-linear relationships and large datasets, the Random Forest (RF) model was employed as the primary predictive tool. Machine learning combined with descriptive statistics and Pearson's correlation was used to describe relationships between nomophobia and other mental health outcomes. Results showed that the severity of nomophobia had significant positive correlations with anxiety, stress, and depression. The RF model demonstrated excellent predictability with a value of 0.9773 for <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex>, showing that it explained about 98 % of the variation in the nomophobia severity scores. This was 15 % better than traditional models such as Logistic Regression (LR) and Support Vector Machines (SVM). This makes the model a dependable tool for the early detection. The feature importance analysis further identifies the key behavioral and psychological predictors of nomophobia and provides some deeper insights into the underlying determinants.