September 20, 2024Journal of Clinical Pathology

Extraction and classification of structured data from unstructured hepatobiliary pathology reports using large language models: a feasibility study compared with rules-based natural language processing

View Full Paper

Ask AI

Bookmark

Authors

RGRuben GeevargheseMemorial Sloan Kettering Cancer Center CSCarlie SigelMemorial Sloan Kettering Cancer Center JCJohn CadleyMemorial Sloan Kettering Cancer Center

Discussion

Member takes

Implication

Retrospective study shows large language models extract pathology elements with up to 94.8% accuracy in hepatocellular carcinoma, indicating potential to streamline oncology data curation.

Key Points

Large language models matched regular expressions by extracting unstructured hepatobiliary data with high accuracy ranging from 84.1% to 94.8% across 88 pathology reports.
Retrospective study of hepatocellular carcinoma cases compared GPT-4 and GPT-3.5 turbo against rules-based natural language processing across five targeted report elements.
Automated extraction using large language models may bypass manual curation and improve structured reporting, significantly accelerating the overall pace of cancer research.

Cite This Study

Geevarghese et al. (2024) studied this question.

synapsesocial.com/papers/68e57c2ab6db64358751bad0 https://doi.org/10.1136/jcp-2024-209669

View Full Paper

Ask AI

Bookmark

Implication

Retrospective study shows large language models extract pathology elements with up to 94.8% accuracy in hepatocellular carcinoma, indicating potential to streamline oncology data curation.

Extraction and classification of structured data from unstructured hepatobiliary pathology reports using large language models: a feasibility study compared with rules-based natural language processing

Discussion

Member takes

Implication

Key Points

Cite This Study

Extraction and classification of structured data from unstructured hepatobiliary pathology reports using large language models: a feasibility study compared with rules-based natural language processing

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider