Tracing evidence¶
This walkthrough connects the NosoGraph data model to the local inspection path. It is about how to review a stored relationship, not how to derive a medical conclusion. The practical sequence is:
condition -> typed claim -> evidence -> study or source -> provenance
Start with a condition¶
Use a repository-backed condition in the local dashboard or the API. The SLE walkthrough provides a compact software example; it does not assert a clinical finding.
Open the typed claim¶
From a condition view, select a claim and note its subject, predicate, and object. The predicate is the relationship type. Read Claims if the structure is unfamiliar.
Inspect evidence direction¶
Open the claim in Evidence Explorer. Review the evidence groups using the canonical labels:
SUPPORTS— the record is represented as supporting the claim.CONTRADICTS— the record is represented as disagreeing with the claim.INCONCLUSIVE— the available records do not establish a clear direction.UNASSERTED— no directional evidence is recorded.
These labels describe stored evidence semantics. They are not proof, a universal confidence score, or causality.
Follow study and source context¶
Inspect source-native identifiers, study metadata, dates, and URLs where present. A missing field remains missing; do not infer a negative finding from an unavailable study or source field.
Read provenance¶
Follow the available provenance details for source identity, snapshot/version, import context, and fingerprint. Provenance explains what those fields support: traceability and reproducibility checks, not scientific certification.
Continue to comparison¶
When the research question is cross-condition, open Compare. Keep KNOWN_ABSENT distinct from NOT_RECORDED: explicit negation is not the same as no recorded assertion.
Research use only. NosoGraph records computational research artifacts. Associations are not causation, and the walkthrough is not medical advice or clinical decision support.