NLP System for Radiology Report QC Predicts Declining Error Rates in Real-World Evaluation

An NLP-assisted quality control system identified significant differences in radiology report error rates by physician grade, location, and time of day. ARIMA modeling predicted a declining trend over 3 months (error rates 0.036, 0.031, 0.029).

• Among 215,870 reports, error rates varied by physician grade, particularly in description, unit, orientation, and organ resection errors.
• Error rates also differed significantly by examination location, device type, and working period (morning/afternoon/evening).
• The NLP system was integrated into the clinical workflow for real-time analysis.

Automated summary

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