Key points are not available for this paper at this time.
Abstract Work-related stress among health professionals is increasing due to high workloads, staffing shortages, and emotional strain. Traditional detection methods rely on self-reported data, which are time-consuming and limited in their ability to capture early stress indicators. Natural Language Processing (NLP) offers automated approaches to analyse unstructured text and may support more effective stress detection. This scoping review maps existing NLP methods used to identify work-related stress among health professionals. The review followed the methodology of the Joanna Briggs Institute (JBI) and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR). Systematic searches were conducted in PubMed, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), the Cochrane Library, the Association for Computing Machinery (ACM) Digital Library, and Institute of Electrical and Electronics Engineers (IEEE) Xplore, covering the years 2013–2024. Six studies met the inclusion criteria. The majority used mixed-method and retrospective designs, applying topic modelling such as Latent Dirichlet Allocation together with thematic analysis. Commonly identified indicators included psychological distress, substance use, and work-related factors linked to suicide. While NLP shows potential for extracting stress-relevant information, current approaches are limited by retrospective data use and lack of real-time clinical integration. Future work should prioritise anonymous, domain-specific text datasets and evaluate NLP tools in real healthcare settings.
Ikae et al. (Thu,) studied this question.