Better evidence for responsible safety research.

Pharmacovigilance & Medicines Safety

A research programme evaluating the quality of medicines-safety information and the responsible use of AI to assist safety-signal prioritisation.

This programme forms part of the unit’s launch agenda. Each project proceeds through defined scientific, resourcing, data-access and governance gates before research execution.

A defined question. A transparent method.

Assess the provenance, completeness and limitations of suitable safety datasets.

Explore methods for prioritising potential signals for qualified human assessment.

Evaluate false positives, missed signals, bias and reproducibility.

Define human review requirements and the boundaries of any future tool use.

AI-assisted signal-prioritisation evaluation

Can an AI-assisted approach support expert prioritisation of potential safety signals compared with a predefined baseline, without obscuring uncertainty or human accountability?

A scoped evaluation using suitable public or properly authorised data, predefined endpoints and expert review. Data permissions and any required ethics or governance review precede analysis.

State what the evidence can support.

This is a research programme, not an operational adverse-event reporting service or a validated safety-monitoring system. Potential signals require qualified assessment; they do not establish causation.

Each project should define oversight, data permissions, any required approvals, analysis methods, validation needs and publication arrangements. Completed results and confirmed partnerships will be identified in the unit’s publication and collaboration records.