Graduate students face significant mental health challenges, prompting higher education institutions to seek innovative solutions for stress management and improved well-being. This study presents a hybrid framework combining wearable technology, physiological biomarkers, and advanced machine learning techniques to monitor and enhance students' mental health. Wearable devices are employed to collect physiological data such as electrodermal activity (EDA), metabolic equivalent (MET), pulse rate, respiratory rate, actigraphy counts, and temperature. These biomarkers serve as critical indicators for assessing stress and overall mental health states. Machine learning algorithms analyze the collected data, enabling accurate assessments and predictive insights into students' well-being.

Additionally, the framework incorporates large language models (LLMs) to process self-reported data from structured prompts or questionnaires. This integration allows for a more precise interpretation of subjective inputs, producing personalized mental health support recommendations. By combining LLM insights with biomarker analysis, the framework adapts to individual needs while maintaining objectivity. This holistic approach enhances the framework's ability to address diverse mental health challenges among graduate students.

The proposed framework demonstrates how stress can be predicted using biomarkers, underscoring their importance in developing and refining mental health interventions. It also highlights the transformative potential of AI-driven solutions in fostering healthier academic environments through data-driven approaches. The study discusses strategic applications of these emerging technologies in educational settings, aiming to support mental health and improve the overall quality of life for students.

Keywords: Mental health, wearable technology, biomarkers, machine learning, large language models, stress prediction, life quality, graduate students, higher education, mental health interventions.