Advanced large language models like GPT-4 accurately identify thoracic diseases in chest CT reports, enhancing pre-operative surgical planning.
Key Details
- 1Five LLMs (GPT-4, Claude-3.5, Qwen-Max, GPT-3.5-Turbo, Gemini-Pro) compared using 13,489 real-world chest CT reports.
- 2GPT-4 achieved up to 75% accuracy in identifying 13 common chest diseases with multiple-choice prompts.
- 3Multiple-choice prompts significantly improved model accuracy compared to open-ended questions.
- 4Fine-tuning GPT-3.5-Turbo increased its accuracy from 42% to 65% in challenging cases.
- 5No single LLM was best for all diseases, suggesting a tailored approach may be optimal.
- 6Future research will use explainable AI tools to increase transparency and reliability.
Why It Matters
The study demonstrates that modern LLMs can act as accurate 'second readers' for radiology reports, possibly reducing diagnostic errors and alleviating radiologist workload. Fine-tuning and prompt design further boost performance, potentially making AI support accessible even in resource-limited settings.
Source
EurekAlert
Related News
•EurekAlert
Study Questions Universal Benefit of AI Virtual Staining in Medical Imaging
University of Illinois researchers found AI-based virtual staining sometimes reduces information utility in medical images, especially with high-capacity networks.
•EurekAlert
Advances in Multimodal Imaging and AI for Radiation-Induced Brain Injury
A state-of-the-art review highlights the use of multimodal imaging and AI to improve diagnosis and management of radiation-induced brain injury (RIBI).
•EurekAlert
Cellular Mechanisms Behind Retinal Oscillations in Night Blindness
Loss of the TRPM1 ion channel leads to rhythmic retinal signals linked to night blindness and other degenerative eye diseases.