How does Burna AI compute adverse event attribution?

For each adverse event, Burna AI assesses every drug in the regimen separately with two published causality methods, WHO-UMC and the Kramer algorithm, and shows its working. For each drug it reports the WHO-UMC category (certain, probable/likely, possible, unlikely, conditional/unclassified or unassessable/unclassifiable) with the criteria that were met, a Kramer category (definite, probable, possible or unlikely), the time from drug start to onset and whether it fits the drug's known effects, the alternative causes it weighed, such as the disease or another medication, and a written rationale. The drugs are then ranked by likelihood. The investigator reviews, edits if needed and signs. Attribution is accepted separately from the grade, so a clinician can agree with the severity and still change the causality call.