Skip to content

Product review method

The model should not be approved as one large document. Reviewers verify individual objects, behaviours, rules, and scenarios.

  • The definition names one real clinical concept.
  • The identity remains meaningful over time.
  • Fields are intrinsic domain information rather than implementation convenience.
  • Relationships and cardinalities match real clinical work.
  • Behaviours use recognizable business intent.
  • Rules cover the important invalid cases.
  • Outcomes describe meaningful facts.
  • Corrections and historical reconstruction are sufficient.
  • Privacy and blinding boundaries are correct.
  • The object is not an unnecessary duplicate of another concept.
  • The correct object or policy owns the behaviour.
  • Required inputs are domain objects or values.
  • Preconditions are complete.
  • Successful outcome is unambiguous.
  • Refusal reasons are meaningful to the business.
  • Authority and evidence requirements are clear.
  • Repeated, corrected, withdrawn, and superseded cases are explained.
  • The behaviour does not rely on a database, API, or UI concept.
  • Every step uses a named object and behaviour.
  • Intended, planned, actual, observed, and decided truth remain distinguishable.
  • No step silently changes several owners at once.
  • Failed, delayed, partial, and conflicting cases are representable.
  • The complete journey is reconstructable from outcomes and evidence.

For each claim, record:

Claim:
Reviewer:
Decision: verified | verified with limits | disputed | deferred
Supported scenarios:
Counterexample or missing case:
Required change:
Rationale:

The purpose is not unanimous terminology. It is explicit, testable product meaning.