AI-based approach to burnout identification from textual data

TL;DR

This work addresses detecting burnout from Russian-language textual data using an NLP approach. It fine-tunes RuBERT for burnout detection on a mixed dataset comprising synthetic sentences generated by ChatGPT and real YouTube comments, with labeling anchored to ICD-11 and standard burnout inventories. The study reports strong performance (accuracy 0.940, AUC-ROC 0.980) and provides a transparent labeling protocol to support reproducibility, while acknowledging the need for external validation. The method offers a scalable tool for monitoring burnout signals in high-stress workplaces, though generalizability across domains remains to be demonstrated.

Abstract

This study introduces an AI-based methodology that utilizes natural language processing (NLP) to detect burnout from textual data. The approach relies on a RuBERT model originally trained for sentiment analysis and subsequently fine-tuned for burnout detection using two data sources: synthetic sentences generated with ChatGPT and user comments collected from Russian YouTube videos about burnout. The resulting model assigns a burnout probability to input texts and can be applied to process large volumes of written communication for monitoring burnout-related language signals in high-stress work environments.

Figures (2)

• Figure 1: Training and evaluation metrics
• Figure 2: ROC Curve