Definition

The application of statistical and computational techniques to large safety databases to identify potential safety signals automatically.

Example

Bayesian data mining algorithms (BCPNN, MGPS) scan EudraVigilance nightly to identify new drug-event combinations with unexpectedly high reporting frequency.

Regulatory source

EU GVP Module IX, FDA AERS

Common mistakes

Treating data mining results as definitive signals rather than hypotheses requiring clinical validation.

Inspector question

What data mining methods do you use? How do you validate data mining outputs?

Related terms