ARCHIVES

College students involved in entrepreneurship are under a considerable amount of psychological strain, indicating a need for strong computational techniques to measure entrepreneurship's psychological effect. A mental health inference structure that uses affective signals from facial feature assessment through deep learning constitutes the core of this research effort. This deep learning structure is established on optimized methods of feature embedding, and a stage-based hierarchical neural processing structure enables the development of the high-dimensional correlation model between emotional attributes and mental health indicators. As previously mentioned, the initial evaluation of the proposed architecture was completed using the FER-2013 dataset created by Kaggle, which is a standard dataset for assessing different affective emotions as represented in different ages and genders. The proposed architecture has demonstrated substantial improvements in classification performance when compared to many alternative models. The proposed structure achieved the highest precision and F1-scores for each of the seven emotional categories that relate to mental health screening. The proposed structure has also shown the ability to generalize well in many of the same emotional categories as the baseline models. However, additional model building experiments established that every component of the proposed architecture contributes to the overall predictive ability of the proposed mental health inference structure. The results of this research demonstrate the potential of using deep learning models centered around affective emotion as a scalable technique for rapid mental well-being