2025 ASEE Annual Conference & Exposition

Mental health challenges among students, particularly within graduate programs, have become a pressing concern for higher education institutions. To address this, a hybrid framework has been developed, integrating wearable technology, physiological biomarkers, and advanced Machine Learning (ML) techniques to monitor and enhance students' stress management and overall quality of life. This framework leverages wearable devices to collect critical physiological data, including electrodermal activity (EDA), metabolic equivalent (MET), pulse rate, respiratory rate, actigraphy counts, and temperature. The above biomarkers are crucial markers to assess stress and mental health states. ML algorithms employed to analyze data can offer precise assessments and predictive insights into students' well-being.
In parallel, Large Language Models (LLMs) are incorporated to analyze self-reported data from students, including responses to structured prompts and/or questionnaires. This allows the system to interpret subjective input with greater accuracy while subsequently generating personalized mental health support recommendations. The integration of LLMs with biomarker data can significantly enhance the framework’s adaptability and objectivity, thus enabling it to more effectively address students’ individual needs. By combining these technologies, the framework can facilitate healthier academic environments using data-driven interventions.
This study demonstrates the model's ability to predict stress by highlighting the important role of biomarkers in developing and enhancing mental health interventions. Additionally, the paper discusses the implications of implementing AI-driven solutions in educational settings, offering a strategic perspective on how emerging technologies can be applied to improve mental health support systems within academic settings.