It means every grade traces to the line in the medical record that produced it. Burna AI attaches to each suggestion the exact source sentence, the CTCAE criterion it meets, the rationale and the attribution reasoning, and records each cited passage with the note it came from and its position in that note. The citation is part of the output, built into the pipeline rather than added afterwards: a grade whose citation set is incomplete is flagged before it is signed, and the signature records the citation set as signed. That serves two readers. The clinician can check the basis of a suggestion independently instead of relying on it, which keeps the decision with the clinician. And a monitor or inspector can trace any filed grade back to its source.
More on ai architecture.
- How is Burna AI's grading engine built?
- Why does Burna AI use twelve agents instead of one model?
- Can a general-purpose LLM grade CTCAE adverse events the way Burna AI does?
- How many patents has Burna AI filed?
- What language models does Burna AI use?
- How does Burna AI prevent AI hallucinations in adverse event grading?