| Abstrakt: |
A report from St. Jude Children's Research Hospital evaluates the accuracy and potential biases of four natural language processing (NLP) tools in labeling chest radiograph reports across different demographic groups. The study, conducted between April 2022 and April 2024, found that while the NLP tools demonstrated high accuracy overall, there were significant biases with poorer performance in older patients. The research highlights the importance of evaluating NLP tools for potential demographic biases in medical imaging applications. [Extracted from the article] |
