Burnout analysis using NLP and medical biomarkers

Abstract

The increasing number of burnout cases has caused people to share their experience on social media, so that they could find support and understanding. Clinical indicators that can determine the risk of burnout are usually performed in psychological tests or by medical biomarkers, but these methods cannot hold the complex spectrum of emotions. This investigation employs a multimodal approach to predict burnout, incorporating natural language processing of social media content alongside physiological biomarker analysis. The goal is to determine better results in the prediction phase, keeping in mind the possibility of detection in the early phases of burnout. The analysis revealed a strong correlation between negative emotions, social media reactions, contrast of the speech and the burnout. It also indicated multiple phases of burnout, from low, medium, to high risk, giving us the possibility of early prediction.