We’re Listening
The ACR is working hard to shepherd and strengthen initiatives designed to translate members’ voices into action.
Read more
Main Content
May Dispatches 2024
May 9, 2024
- Looks at Quality of Large Language Model Responses to RO Patient Care Questions
- Significant Benefits Seen After Five Years of Using Radiology AI ROI Calculator
- ACR Announces Recipients of Bruce J. Hillman Fellowship in Scholarly Publishing
New Study Looks at Quality of Large Language Model Responses to RO Patient Care Questions
JAMA Network Open recently published a study showing that ChatGPT answered questions regarding care with little risk of inaccuracies, suggesting that ChatGPT could be an important resource for radiation oncology (RO) patients. The chatbot was tested by clinicians at Northwestern University in Chicago using common care questions, with ChatGPT responding with responses compared to human experts. The only issue that Amulya Yalamanchili, MD, and her colleagues detected was that the readability was higher than recommended.
The team wanted to verify if ChatGPT can provide promising answers for radiation treatment questions much like it has in other areas such as medical test questions and simplifying radiology reports. ChatGPT was given 115 questions from the group based on Q&As from sites sponsored by numerous organizations such as the ACR and RSNA. Three radiation physicists and three radiation oncologists ranked the responses based on how correct, complete and concise they were compared to answers from experts online using a five-point Likert scale. The team found that ChatGPT scored better or similar in 108 responses for correctness, 89 for completeness and 105 for conciseness.
Significant Benefits Seen After Five Years of Using Radiology AI ROI Calculator
The benefits are positive regarding a calculator for assessing radiology AI, with the return deemed “substantial” after a five-year period following its introduction. The experts and researchers that built this tool aimed to help imaging leaders measure the quantity of comparative costs, estimated revenues and the value of AI platform use within U.S. hospitals. This calculator was built using a review of imaging AI-related literature and expert interviews, per a study published by the JACR®.
AI introduction to radiology workflow in hospitals has led to labor time reductions and delivery of ROI at 451% during this five-year period, with returns increasing to 791% when also keeping radiologist time savings in mind. Time savings for radiologists, based on the calculator, included more than 15 eight-hour workdays of wait time, 78 days in triage, 10 days in reading and 41 days in report time.
ACR Announces Recipients of Bruce J. Hillman Fellowship in Scholarly Publishing
The ACR has awarded the 2024 Bruce J. Hillman Fellowship in Scholarly Publishing to Fatima Elahi, DO, MHA, and Madison Kocher Wulfeck, MD, MBA, CIIP. Elahi is completing her diagnostic radiology residency at the Hospital of the University of Pennsylvania and will embark on a mammography fellowship at the Mayo Clinic in Rochester. Wulfeck is a cardiothoracic radiologist in Florida with Radiology Partners, and previously was on the JACR Trainee Editorial Board.
The Bruce J. Hillman Fellowship in Scholarly Publishing allows its recipients, who are qualified staff radiologists or radiologists-in-training, with focused experience in medical journalism, editing and publishing, to gain more experience with the JACR®. The fellowship is named after the founding editor-in-chief of the JACR, Bruce J. Hillman, MD, FACR, who held this position from the JACR’s launch in 2004 until 2019. Hillman passed away on Jan. 9, 2024.
