Deep Neural Network Framework for Early Detection of Student Depression Based on Academic and Behavioral Data

Depression among university students has become a growing concern due to its negative impact on academic performance, social functioning, and overall well-being. This study aimed to develop and evaluate a Deep Neural Network (DNN) model for the early prediction of depression based on academic, behavioral, and lifestyle variables. The dataset consisted of 502 student records containing attributes such as academic pressure, study satisfaction, sleep duration, dietary habits, financial stress, and family mental health history. Data preprocessing involved label encoding for categorical variables and normalization for numerical features, followed by model training with dropout and L2 regularization to reduce overfitting. The proposed DNN achieved a moderate performance, with an overall accuracy of 59 percent, a macro-averaged F1-score of 0.55, and an Area Under the Curve (AUC) value of 0.64. Although the model demonstrated limited discriminative ability, it successfully generalized across training and validation data, indicating a stable learning process. The findings reveal that academic pressure, sleep duration, and financial stress were among the most influential predictors of depression. These results suggest that while academic and lifestyle factors alone may not fully capture the complexity of mental health conditions, deep learning can still serve as a useful tool for early screening and risk identification. The study highlights the potential of artificial intelligence in supporting student mental health monitoring and promoting preventive interventions in higher education.