How does Burna AI grade CTCAE adverse events?

Burna AI reads a clinical note, a recorded visit or a batch of notes, finds every adverse event in it, and proposes a CTCAE grade for each one against the version the trial names, v5.0 or v6.0. The engine standardizes drug names, resolves each phrase to a CTCAE term and its MedDRA Lowest Level Term code, reads that term's published grade criteria, takes the patient's known comorbidities and medications into account and notes where a comorbidity may explain a symptom, and scores attribution to every drug in the regimen with the WHO-UMC and Kramer methods. Each proposal carries its evidence: the source sentence, the CTCAE criterion it meets, the rationale and the attribution reasoning. A grade whose citation set is incomplete is flagged before signature. A qualified investigator reviews, edits if needed and signs under 21 CFR Part 11, and only then does the grade become a record.