Standards and Compliance

ATLAS is built around the established vocabulary and methods of pharmacovigilance so that its output maps cleanly onto the way safety professionals already work. This page describes the standards ATLAS aligns to and how it handles data and AI responsibly.

Pharmacovigilance standards

• ICH E2A — definitions of seriousness and expedited-reporting concepts inform how ATLAS classifies events.
• ICH E2C (PBRER) / GVP Module VI & IX — the structure of periodic reporting and literature surveillance shapes ATLAS’s drafting and signal workflows.
• ICSR minimum criteria — ATLAS checks each record for the four elements of a valid individual case safety report (identifiable patient, identifiable reporter, suspect product, adverse event).
• WHO-UMC causality — causality assessments use the WHO-UMC categories.
• MedDRA — adverse events are mapped toward MedDRA Preferred Terms where possible. MedDRA is a registered trademark of the IFPMA; ATLAS is not endorsed by or affiliated with the MedDRA MSSO.
• CIOMS — case line listings follow CIOMS-style conventions.
• Disproportionality methods — PRR, ROR, and the chi-squared signal criterion follow standard signal-detection practice.

Data sources

ATLAS retrieves data from PubMed/MEDLINE (NCBI), Europe PMC, and openFDA/FAERS. These sources are used under their respective terms; the underlying records remain the property of their owners.

AI governance

ATLAS uses automated language models for extraction and drafting. Outputs are designed to be traceable — each carries the model and timestamp that produced it — and are presented as decision support for qualified human review, never as autonomous determinations. AI processing is disclosed in our Privacy Policy.

Validation status

ATLAS is a decision-support tool and is not, by itself, a validated computerized system or a regulatory system of record. Organizations that incorporate ATLAS into a regulated process are responsible for any validation their own quality system requires.