Example of epistemic discipline
FAERS — Pharmacovigilance under a truth constraint
What FLORN will establish, what it refutes, and what it refuses to certify on a set of spontaneous reports.
9 findings · 7 established · 2 refuted · 0 causal
No causal certification.
Because the data do not allow it.
Here is what an engine that refuses to lie does with one of the noisiest datasets in pharmacovigilance.
FAERS combines spontaneous reports, reporting bias, and no experimental protocol. Most systems answer with signals or causal graphs of fuzzy confidence.
FLORN does otherwise:
- It certifies 7 descriptive and temporal facts it can actually support.
- It explicitly refutes 2 candidate relations.
- It refuses any direct causal certification.
The certified fact
What this analysis separated
Established 7 descriptive and temporal facts supported by the evidence.
Refuted 2 candidate relations.
Unresolved Any direct causal certification.
On FAERS, FLORN did not invent causality.
That is exactly what it is asked to do.
This is a public example of a FLORN result. It is not medical advice. Only a completed analysis on a signed-in account produces a judgment for your own data.