Performance of a ChatGPT-based ergonomic assessment for desk-working posture in young adults

BackgroundArtificial intelligence (AI)- supported ergonomic assessments identify occupational musculoskeletal risks, though agreement with physiotherapist-based evaluations remains insufficiently studied to date.ObjectiveThis study aimed to investigate the relationship between AI-assisted ergonomic risk analysis and physiotherapist-based ergonomic assessments among young adults performing desk-based work.MethodsThis cross-sectional study included 133 desk workers aged 18-25 years who used electronic devices at their desks for more than 4 h per day. Ergonomic risk was evaluated using two methods: Rapid Upper Limb Assessment (RULA) and OpenAI ChatGPT. Pain intensity was measured with the Visual Analog Scale (VAS). Pearson correlation analysis assessed the relationship between RULA and ChatGPT scores. Receiver Operating Characteristic (ROC) analysis was performed to assess discriminative performance and determine the optimal cutoff value for the AI-based assessment.ResultsThe participants' (48.9% females and 51.1% males) mean age was 24.02 ± 6.34 years. The mean RULA and ChatGPT scores were 4.65 ± 1.31 and 45.71 ± 16.31. A statistically significant positive correlation was observed between ChatGPT and RULA scores (r = 0.304, p < 0.001). ROC analysis showed moderate discriminatory performance for the AI-based assessment (AUC = 0.659; 95% CI: 0.565-0.752; p = 0.002). The optimal cutoff for the ergonomic risk analysis using ChatGPT was 42.5, yielding 72.4% sensitivity and 56.0% specificity.ConclusionsAI-assisted ergonomic risk analysis showed moderate agreement with conventional physiotherapist-based assessments and may serve as a rapid, accessible screening tool for desk workers. Nevertheless, clinical interpretation by healthcare professionals remains essential for accurate ergonomic evaluation and decision-making.