Large language model pilot

We are conducting a pilot of generative artificial intelligence to create human-readable synopses for some of the most relevant anticancer regimens on HemOnc.org. The goal of this work is to further contextualize the regimen information that is currently primarily in structured format, providing a historical perspective and overview for site users.

The following is a general overview of our process:

A commercially available large language model (e.g., Open AI's Chat-GPT; Google's Bard) is prompted to generate a synopsis of a given regimen, using citations from the literature. The typical format is “Create a summary of the evolution of XX regimen for treatment of XX. Use citations from the literature to support this”. The page and/or section editor then reviews and substantially edits the summary for clarity, readability, and accuracy. The synopses are reviewed and approved by the Editor-in-Chief or Deputy Editor before posting on the website.

Any synopsis that has been created in this manner will have a footnote with a pointer to this page, along with the date that it was created.

Please note that our general disclaimer applies to all content on HemOnc.org, regardless of its source.

We are seeking your feedback on this pilot! Please fill out our REDCap form or send us an email at contact @ hemonc.org.

This work is supported in part by NCI U24 CA265879.