Evaluation Model of the Mental Health Education Effectiveness Based on Deep Neural Networks

Abstract

This research develops a deep neural network model called DNN-MHE to evaluate mental health education effects. A questionnaire survey collected data on 916 students' mental health knowledge, attitudes, and behaviors. DNN-MHE uses five fully connected layers to predict mental health metrics. Experiments demonstrate that DNN-MHE achieves 99.46% accuracy, outperforming RNN, CNN, and shallow MLP models. Ablation studies validate the impact of training iterations, number of neurons, and number of data samples on performance. Overall, DNN-MHE enables accurate and efficient analysis of mental health education with practical implications for improving university programs.