AI-Assisted Radiographers: High Sensitivity, Improved Specificity in Acute MSK X-rays

AI-supported radiographers interpreting acute MSK X-rays had 94% sensitivity and 86% specificity, improving on AI alone's specificity (71%) but with slight sensitivity drop (96%). Variability and suboptimal specificity highlight need for training.

• Compared with AI alone (96% sensitivity, 71% specificity), AI-supported radiographers showed slightly lower sensitivity (94%) but markedly higher specificity (86%).
• Performance varied by radiographer; those with ≥5 years of experience had higher sensitivity (p=0.01) but lower specificity (p=0.02) than less experienced colleagues.
• Suboptimal specificity suggests room for improvement via training or workflow refinement.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.