AI-Based CT Queue Prioritization in the ED: A Simulation Study

A LightGBM model was trained on 313,966 ED visits to predict CT clinical actionability. In a discrete-event simulation of 20,795 studies, an AI-driven prioritization policy was compared to first-in-first-out to assess wait times.

• The discrete-event simulation was calibrated to real-world ED operations.
• Data spanned August 2020 to August 2024, with the test set from March to August 2024.

Automated summary

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