FDA: AI-augmented signal detection tool validated

In a development closely watched by the pharmacovigilance community, the FDA's Center for Drug Evaluation and Research disclosed that it has internally validated an AI-augmented signal detection tool for use alongside traditional disproportionality analysis methods. The tool uses natural language processing to extract safety-relevant information from FAERS case narratives and integrates it with structured data for enhanced signal prioritization.

The validation study compared the AI-augmented approach against standard MGPS analysis across a set of 50 known positive and negative drug-event associations. The AI-augmented tool demonstrated improved sensitivity without a corresponding increase in false-positive signals.

What this means for practice: This is an internal FDA capability, not a requirement for industry. However, it signals the direction of regulatory expectation — marketing authorization holders should anticipate that regulatory agencies will increasingly use AI-enhanced methods for signal evaluation, raising the bar for industry signal management programs.

EMA: revised GVP Module IX

The European Medicines Agency published its long-anticipated revision of Good Pharmacovigilance Practices Module IX on signal management. The revision incorporates several methodological updates including recognition of Bayesian shrinkage methods alongside frequentist disproportionality analysis, guidance on integrating real-world data sources into signal evaluation, and updated timelines for signal assessment procedures.

TGA: compounded semaglutide safety alert

Australia's Therapeutic Goods Administration issued a safety alert regarding compounded semaglutide preparations following reports of serious adverse events related to dosing errors. The alert identified cases where compounded preparations contained incorrect concentrations of semaglutide, leading to significant hypoglycemia and gastrointestinal adverse events.