Design and Development of Ensemble Deep Learning Framework for Psychological Health Assessment of College Students

The college students' psychological condition is a growing concern with increasing academic stress, social pressure, and lifestyles. The standard methods of assessing mental health, e.g., clinical interviews and questionnaires, are typically plagued by the issues of scalability, real-time data monitoring, and subjective interpretation. In the recent times, machine learning and deep learning algorithms has attracted the brighter light of deployment in designing the intelligent assessment systems for students’ psychological health conditions. However, computational overhead in these models bottlenecks the system design to achieve the high assessment performance. To solve this challenge, this research article proposes a novel gated recurrent network ensemble with the feed forward networks (FFDN) for evaluating college students' psychological well-being using various behavioural and emotional cues. Extensive experiments are conducted using Kaggle datasets, and performance metrics including accuracy, precision, recall, specificity, and F1-score are calculated and compared across various deep learning models. Finally, the prominence of the recommended approach is shown by comparing with the varied residing learning approach. Experimental outcomes also demonstrates that the recommended learning model has surpassed the varies residing model in producing the 97.3% of accuracy, 96.9% of precision, 96.5% of recall, and 96.6% of F1-score respectively.