Improving clinical text classification using large language models guided by semantic knowledge

Current state of the art models for performing clinical text analysis do not yet represent technologies that can be incorporated into tools for live use by medical professionals and practitioners in hospitals due to the discrepancy between the data used in research and the data created in and used by hospitals. Public datasets utilized by natural language processing (NLP) research groups are heavily processed before use in research both by necessity (removal of sensitive personal information) and to improve the ability of language-processing models to extract information.