Abstract: The increasing pressure of academic life significantly affects students mental health, making early detection of stress essential to prevent long-term consequences. Extended exposure to academic pressures can negatively impact students' emotional health and impede their academic development. This research presents a system aimed at recognizing early signs of stress in students prior to any decline in their mental health. Methodology utilizes a mix of machine learning algorithms and analysis of multimodal data. We examine audio recordings through Natural Language Processing (NLP) techniques, concentrating on identifying stressed and not stressed words to assess emotional tone and stress indicators derived from speech. Visual information, obtained through student photographs, is analyzed by a Convolutional Neural Network (CNN) to identify subtle facial expressions linked to stress. Additionally, student responses to structured questionnaires are examined using a Random Forest algorithm to identify behavioral patterns linked to stress. By integrating insights from audio, visual, and questionnaire data, the system enhances accuracy in stress prediction across various academic settings. This tool can help educational institutions track student well-being, facilitating prompt interventions to foster a healthier learning atmosphere.
Keywords: Facial Expression Recognition, Audio Analysis, Natural Language Processing (NLP), Stress Prediction, Image-Based Stress Analysis.
|
DOI:
10.17148/IJARCCE.2024.131111
