Predicting the Impact of Social–Emotional Skills on Student Well-Being and Health in Higher Education Using CNN and LSTM Models

作者：G. Vasanti, A. Parasar, S. Pragadeeswaran, M. Y. Sayed, Yashwant Waykar, Anvesh Perada · 发表于：2025 5th Asian Conference on Innovation in Technology (ASIANCON) · 年份：2025 · DOI：10.1109/asiancon66527.2025.11280617

Evidence suggests that implementing social and emotional learning (SEL) practices into the classroom improves student confidence, performance in the classroom, cognitive development, and teacher-student interactions. Improving motivation, social awareness, academic achievement, and classroom engagement is possible by prioritising student well-being in the context of Health in HE. College campuses are becoming increasingly diverse in terms of student demographics, academic performance, and life experiences. Consequently, there is an urgent need to cultivate welcoming classroom settings that promote the development of social and emotional competencies as well as resilience. Data preprocessing, feature selection, and training are the three stages of this research's model, which aims to improve students' mental health and academic performance. After applying a min-max scaling strategy to normalise the dataset, the feature selection process known as mRMR is utilised. The LSTCN model is an innovative combination of CNN and LSTM, two well-known deep learning techniques. In order for LSTM to provide a final prediction, CNN is employed to extract sample features. When compared to standalone CNN and LSTM models, the LSTCN model achieves the best prediction accuracy of 94.71%, suggesting that it is useful for promoting AI-supported strategies to enhance HE students' happiness and performance.