QEM

The Qualifying Evidence Matrix (QEM) standard for verifiable evidence is used for the QuantBayes evidence matrix layer. It records evidence outcomes for item-rule pairs and converts them into the binary matrix used by QuantBayes.

The standard can be found at: https://www.swissgenomicsassociation.ch/pages/sga_qem/

The QEM standard defines a minimal and interoperable way to encode verifiable evidence derived from rule-based evaluations. QEM is designed for situations where analysis results may be produced by proprietary or opaque algorithms, but where the availability of underlying evidence must remain auditable, reproducible, and independently verifiable.

Raw outcomes

Each item-rule pair has one raw outcome:

TRUE
FALSE
NA

Canonical mapping

TRUE  = problem found    = 0
FALSE = evidence present = 1
NA    = not available    = 0

Why FALSE can mean evidence present

Rules are written as failure checks. The rule asks whether a problem, failure, absence, contradiction, or missing evidence condition is present.

Therefore:

TRUE means the problem was found.
FALSE means the problem was not found, so evidence is present.
NA means the check cannot be assessed.

Matrix values

The QEM matrix contains only:

0
1

1 means evidence is present under the selected rule.

0 means evidence is not counted as present.

Interpretation boundary

QuantBayes measures evidence sufficiency under a declared evidence profile. It does not directly measure truth, pathogenicity, safety, or actionability.