Reliability and agreement during the Rapid Entire Body Assessment: Comparing rater expertise and artificial intelligence.
Authors: Balogh Denise, Cui Xiaoxiao, Mayer Monique, Koehncke Niels, Dueck Ryan, Lang Angelica E
Journal: PloS one
Summary
# Editorial Summary Ergonomic assessment through the Rapid Entire Body Assessment (REBA) is increasingly recognised as a tool for identifying workplace injury risk in equine professionals, yet its reliability depends heavily on rater experience—a limitation that artificial intelligence may help address. Researchers compared REBA scoring consistency across novice, intermediate, and expert human raters against TuMeke Risk Suite (an AI ergonomic assessment software) when analysing 21 videos of veterinarians performing equine radiography procedures. The AI demonstrated perfect intra-rater reliability (ICC = 1.0), substantially outperforming novice raters (ICC = 0.51) and matching expert performance more closely (ICC = 0.89), though agreement between AI and expert raters varied significantly by body region—strongest for trunk, legs, and upper arm positioning but weakest for neck, wrist, and lower arm assessment. For practitioners and clinic managers without access to movement specialists or experienced assessors, integrating AI-assisted REBA analysis could standardise injury risk identification and help guide ergonomic interventions, particularly when evaluating large-muscle-group postures; however, the technology's limitations with distal upper limb positioning suggest human expert review remains valuable for comprehensive assessment.
Read the full abstract on PubMed
Practical Takeaways
- •This study addresses human ergonomics during equine work rather than equine health or performance — limited direct application to equine-focused practitioners
- •Findings suggest AI ergonomic tools could standardize workplace safety assessments in veterinary clinics, potentially reducing musculoskeletal injury risk among staff
- •If your clinic performs repetitive radiography or other detailed work, AI-assisted ergonomic assessment may provide more consistent evaluation than relying on individual rater expertise
Key Findings
- •TuMeke Risk Suite AI achieved perfect intra-rater reliability (ICC = 1.0) compared to expert rater (ICC = 0.89) and novice rater (ICC = 0.51) on final REBA scores
- •Agreement between expert raters and AI software was highest for trunk, leg, and upper arm assessment but lowest for neck, wrist, and lower arm
- •AI-assisted REBA assessment may improve reliability for less experienced users in ergonomic evaluations of equine veterinary work