One value
A performance score is present.
Encoding uncertainty · Entry 020
A blank field is not a no.
A sparse assessment and a complete negative looked identical to the first engine. One explicit review-scope field became the permission required before software could emit a confident negative.
Chapter 01
One observed fact could not describe the review.
A synthetic score with every other field blank might reflect a complete assessment or a failed extraction. The facts alone cannot tell those states apart.
Many blanks
No provenance explains why.
RED?
A confident negative would manufacture certainty.
Chapter 02
Scope must authorize the negative.
PARTIAL is the safe default. It returns YELLOW with review_incomplete. Only a defined COMPLETE review permits the rules to consider RED.
Same facts. Different provenance.
Different permission to claim.
Different permission to claim.
Chapter 03
There are two kinds of “we do not know.”
Missing means an applicable field has no usable value. Not applicable means the criterion does not belong to this clinical context.
Missing
Expected for the assessment, absent or unusable.
Not applicable
Irrelevant to the selected diagnosis or rule family.
Why it matters
Completeness and user-facing explanations remain honest.
Chapter 04
The twin test pins the authority boundary.
The synthetic facts remain identical. Only review scope changes what the engine is permitted to say.
PARTIAL
YELLOW · review incomplete
Scope gate
Completeness is explicit provenance, not inference.
COMPLETE
Configured rules may evaluate a negative band.
Clinical boundary
Bands describe documentation, not eligibility.
CMS states that Medicare hospice certification depends on physician clinical judgment and current clinically relevant information. CareFlow organizes evidence; it does not certify terminal illness.
Visual edition
A blank field is not a no.
A narrated walkthrough of the sparse assessment, review-scope gate, missing versus not applicable states, twin test, and clinician authority boundary.