Detecting Nurse Burnout with Multi-Modal Large Language Models: A Feasibility Study

Occupational burnout among psychiatric nurses poses a significant challenge to mental healthcare systems. This research applies Multi-modal Large Language Models (MLLMs) to assess burnout in real-time by combining data from facial expressions, speech, and text. The study evaluates the effectiveness of sentiment analysis and multi-modal emotion recognition (MER) in detecting early signs of burnout. A total of 93 psychiatric nurses (75.27 % female) participated in the study. The results revealed significant correlations between sentiment analysis and burnout labels, with a chi-square test yielding <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{p}$</tex>-values of 0.039 for image analysis and 0.001 for word-based analysis. Despite its promising results, future research warrants the application of diverse models to further refine the system's predictive capabilities.