Dennis Thompson HealthDay Reporter
health day
TUESDAY, April 16, 2024 (HealthDay News) — Artificial intelligence can act as a useful backup editor for radiologists, ensuring their reports are accurate and reliable, new research says. announced.
The OpenAI program GPT-4 performed nearly as well as a group of radiologists at detecting the errors described in the report, according to the study published April 16 in the same journal. Radiology.
“This efficiency in error detection means that AI can help optimize workflows within radiology departments, ensuring that reports are accurate and quickly available, and enhancing radiology departments’ ability to provide timely and reliable diagnoses. “This may suggest a future in which research will be carried out,” said the lead researcher, Dr. Roman Goertz, a radiology resident at Cologne University Hospital in Germany.
This is the first study to directly compare GPT-4 to human performance reviewing radiology reports for accuracy, speed, and cost-effectiveness, Gertz said.
For the study, Gertz and his colleagues collected 200 radiology reports based on X-rays, CT scans, and MRI images at a single hospital between June and December 2023.
The researchers intentionally inserted 150 errors into the 100 reports, including omissions, misspellings, confusing sentences, and other errors.
They then asked both GPT-4 and a group of six radiologists (two senior radiologists, two attending physicians, and two residents) to independently review the report and spot any errors. I requested it.
Overall, the GPT-4 had an 83% error detection rate, compared with 89% for senior radiologists and 80% for attending radiologists and radiology residents, the researchers said. .
There was no evidence of a significant difference in mean performance on the GPT-4 compared with radiologists.
The AI program also had faster processing times per report than the fastest human readers and lower average revision costs per report than the most cost-effective radiologists.
The study shows that AI can improve patient care by increasing the accuracy of radiology reports and the speed of report response, Gertz said.
“This research addresses critical health care challenges, including increasing demand for radiology services and pressure to reduce operating costs,” Gertz said in a journal news release.
“Ultimately, our research shows how AI can increase efficiency, minimize errors, and provide broader access to reliable and affordable diagnostic services, especially through applications like GPT-4. “This is a fundamental step to improving patient care outcomes.” added.
Source: Radiological Society of North America, News Release, April 16, 2024
Copyright © 2024 HealthDay. All rights reserved.
