Definition

Use of machine learning or statistical algorithms to identify potential safety signals from large datasets, with human oversight. EMA and FDA expect documented validation and transparency.

Example

Disproportionality analysis augmented by NLP on narrative text to flag drug–event combinations.

Regulatory source

EMA Big Data Steering Group, FDA AI/ML Discussion Paper 2024–2026

Common mistakes

Relying on AI output without medical review or without documenting algorithm limitations.

Inspector question

How do you validate and document AI tools used in signal detection?

Related terms