CIOMS — Ethics, Harmonisation, and the AI Framework

The Council for International Organizations of Medical Sciences is a non-governmental organisation established jointly by WHO and UNESCO in 1949. In pharmacovigilance, CIOMS working groups bridge scientific evidence, regulatory expectation, and ethical principle — defining international consensus in areas formal regulation hasn’t fully specified.

Several CIOMS working groups shape daily PV work: WG VI covers SUSAR reporting from clinical trials, WG VIII covers additional risk minimisation measures, WG IX covers signal detection methods, WG X covers evidence synthesis across trials, and WG XII underpins the benefit-risk sections of the PBRER.

WG XIV — finalized December 2025 — is the first comprehensive international AI governance framework for the discipline, and establishes seven operating principles: transparency (no black-box models — explainable AI required), validation (against representative datasets, with documented precision/recall/F1 metrics), human oversight (a named human reviewer for every AI output before regulatory action), bias testing (demographic, geographic, and clinical bias must be tested and documented), audit trails (every AI action reproducible and attributable, with model version control), accountability (the QPPV retains final legal and ethical responsibility — this cannot be delegated to a technology vendor), and ongoing monitoring (model drift requires detection and response protocols).

Notice how closely WG XIV’s seven principles mirror the ethical foundations covered in Module 1 — respect for persons becomes bias testing and human oversight; beneficence and non-maleficence become validation and ongoing monitoring; and accountability is, in essence, a direct descendant of the same "who is responsible" question the Belmont Report answered for human research seventy years earlier. AI governance in PV isn’t a new ethical framework invented from scratch — it’s the same research-ethics lineage applied to a new kind of decision-maker.