Topline
ChatGPT may be able to respond to questions about radiation oncology as well as trained professionals or better, which could reduce clinician workload and burnout, according to a small study published Tuesday, the latest findings suggesting AI could improve treatment and diagnosis of cancers.
Key Facts
ChatGPT’s responses to 115 common questions about radiation oncology—radiotherapy or chemotherapy—were on par with or exceeded answers provided by experts on the internet, often explaining complex topics in simpler terms, according to a Northwestern Medicine study published Tuesday in JAMA Network.
Researchers compiled questions and responses about radiation oncology from websites affiliated with the National Cancer Institute and the Radiological Society of North America before using those questions as prompts for ChatGPT.
Three radiation oncologists and three radiation physicists then scored each response for accuracy, completeness and conciseness when compared to responses from online experts.
Amulya Yalamanchili, a radiation oncology resident at Northwestern Medicine, said the technology could help reduce workload and burnout among physicians, who are increasingly more in demand as cancer cases rise.
Though ChatGPT provided reliable answers, several responses were concerning because they used language at a college reading level, including complex medical jargon, researchers warned.
What To Watch For
The researchers and other Northwestern University scientists are developing a generative AI app to help answer questions from patients about radiation oncology treatment, researchers said.
Big Number
650,000. That’s how many patients with cancer receive a form of radiation oncology treatment in the U.S. each year, according to a 2022 study. About 60% of patients diagnosed with a form of cancer require radiotherapy or chemotherapy.
Key Background
Medical professionals have increasingly suggested AI could benefit oncology as the technology quickly develops. Physicians from Stanford University wrote last year that AI could be used as a tool to help reduce errors in diagnosis and improve the cancer detection rate, though the technology was still “years away” from helping predict treatment outcomes and cancer screenings. Using AI for cancer prognosis, diagnosis and prediction may “one day” result in more effective cancer treatment, according to a 2023 study. Though AI may eventually be beneficial in oncology, there are concerns about “unique ethic and legal considerations,” including “inherent” biases in data sets used to train the technology.
Surprising Fact
Researchers from the University of Oxford and the University of Manchester used AI to identify a second form of prostate cancer earlier this month. AI was used to study changes in the DNA of prostate cancer samples from 159 patients, which indicated two distinct cancer groups among those patients, researchers said.