Congratulations to our graduate student Ethan Stolen and Assistant Professor of Radiation Oncology James Sohn, who have just published a new paper in the Journal of Applied Clinical Medical Physics (JACMP). This high-impact clinical journal has an international audience of medical physicists committed to improving clinical practice through innovation and research. They were joined by researchers at the UChicago Data Science Institute, the University of Wisconsin-Madison, and Virginia Commonwealth University, demonstrating their commitment to improving radiation therapy at other institutions around the world.
Their paper represents a novel application of artificial intelligence to analyze the reliability of the linear accelerators (LINACs) that are used daily to treat cancer patients. The work utilized 10 years of extensive maintenance records from UChicago Medicine’s four LINACs. Their work demonstrates how Large Language Models (LLMs) can transform extensive, decade-long unstructured maintenance records into clear, actionable insights regarding equipment failure patterns and downtime causes. This makes their study one of the longest, most complete analyses of LINAC maintenance records.
This data-driven approach directly informs maintenance strategies and inventory management at UChicago Medicine, ensuring higher machine availability and uninterrupted patient care. Their innovative methods highlight UChicago Medicine’s commitment to leveraging advanced data science to enhance the quality and consistency of radiation therapy.
Read the paper here:
Youn, Y., Schulz, J. B., Stolen, E., Farrey, K., Muresan, E., Kim, S., & Sohn, J. J. (2025). Linear accelerator (linac) downtime analysis assisted with a Large Language Model (LLM). Journal of Applied Clinical Medical Physics, 26 e70400. https://doi.org/10.1002/acm2.70400