Skip to content
Blueprint · Healthcare

Explain the rating change. Performance, rule, source and timing.

A worked example for health plan quality teams: why a quality measure moved, whether it was performance or methodology, the gap to the next threshold and when an improvement can count. Operational analysis only, nothing clinical.

Healthcare
The question

Why did this measure drop to three stars, and what would it take to reach four?

Inputs

What the agents read.

Quality results
Measure rates, scores and historical performance
Published methodology
Releases, cut points and technical notes, by year
Measure definitions
Specifications and their changes over time
Contract context
Rating year, applicable measures and dependencies
Run

What happens, step by step. Each step is logged with its sources.

  1. 01

    Find the right sources

    Match the measure, contract, publication and rating year.

  2. 02

    Compare the versions

    Spot changes in performance, thresholds, definitions and methodology.

  3. 03

    Run the calculation

    Reproduce the rating and measure the gap to the next threshold.

  4. 04

    Check the timing

    Connect the measurement period with the publication and rating year.

  5. 05

    Ask an expert

    A quality leader checks the evidence, assumptions and likely impact.

Agent Hive · Run 0912
Why did measure M-12 drop to three stars, and what would four take?
  1. ResolvedM-12, contract A, rating year 30.4s
  2. Compareddefinitions, years 2 and 31.0s
  3. Readcut points, year 3 release0.7s
  4. Reproducedrating from the published file0.5s
  5. Wroteexplanation with 5 sources0.4s
Answer4 sources
71%3 stars
Measure M-12, rating year 3

Performance held at 71%. The four-star cut point moved from 70% to 75%, so the rating fell on methodology, not performance.

measure_ratescut points y3technical notesdefinition y3
Ask a follow-up…
Approvals

Where people decide.

  • Needs you

    Interpretation

    A quality leader confirms the interpretation before anyone acts on it.

  • Needs you

    Release

    Only approved versions go live. Every update reruns the same test set.

Outputs

What you get back.

Explanation
What changed, which rule applied and when it can matter
Gap calculation
Reproduced rating and distance to the next threshold
Source trail
Files, assumptions and dependencies behind the answer
Reasoning trail

From a 71% score to the question that matters next.

The reasoning path for one measure: the rate, the cut point that set the rating, the next threshold and the gap, with the timing that decides when it counts.

Fig. 1 · Measure M-12, rating year 3Illustrative data
Measure performance71%Reported rate, unchanged
Current rating3 starsYear 3 cut points applied
Next threshold75%Four-star cut point, year 3
Gap4 ptsCalculated difference
Measurement periodWhen performance is earned
Publication and rating yearWhich methodology applies
What this could affectWhat the team should test next
Performance held at 71%; the four-star cut point moved to 75%. The gap is reproduced from the published file, and the timing shows when an improvement can count.
Data sources

Put the rating and its source side by side.

A measure value means little without the definition, cut point, contract and year behind it. Agent Hive keeps those pieces together.

Fig. 2 · Connected sources
Sources stay in place; Agent Hive reads them through read-only connectors and builds one shared model. Logos are trademarks of their respective owners.
  • Quality results

    Measure rates, scores and history from the quality warehouse.

  • Methodology files

    Published releases, cut points and technical notes.

  • Contract context

    Rating year, applicable measures and dependencies.

Agent workflow

Compare, calculate, explain. An expert confirms.

Version comparison and the rating calculation run side by side. The explanation waits for a quality leader before it is used.

Fig. 3 · Workflow runIllustrative data
The run as the platform draws it: tool steps, specialist agents, joins, branches and the approval it is waiting on. Drag to pan; Ctrl or ⌘ + wheel to zoom.
Versioned evidence

Use the right rule for the right year.

A measure can keep its name while its definition, weight or threshold changes. The quality evidence map keeps measures, contracts, thresholds, publications and years apart.

Fig. 4 · Knowledge modelIllustrative data
The quality evidence map: a measure, its definition in each year, the cut points and publications that apply, and the timing that links them.
Fig. 5 · Provenance of one valueIllustrative data

Provenance trail. Provenance of “on-time rate = 0.874”: on-time rate = 0.874 ← query run ← metric definition ← delivery policy §4 (read right to left: the fact was derived from each step in turn). Every step: entity “on-time rate = 0.874” (the fact in the answer); wasGeneratedBy query run. activity “query run” (2026-10-21 · 09:14); wasAssociatedWith analyst agent, used orders dataset, used metric definition. agent “analyst agent” (ran the query). entity “orders dataset” (snapshot 2026-10-20); wasDerivedFrom ERP export. entity “metric definition” (v3 of 3), versions v1 first draft; v2 excludes cancelled lines; v3 late = after promised date; wasDerivedFrom delivery policy §4, wasAttributedTo data steward. entity “ERP export” (nightly load). entity “delivery policy §4” (source document). agent “data steward” (approved v3).

The calculation that produced it, the files it used and who approved the interpretation.
Question types

Ask the questions your quality team will ask.

  • Explain the movement

    Why did this measure move?

    Separate a real performance change from a calculation or data change.

  • Check the rule

    Performance or methodology?

    Show what changed in the definition, threshold, weight or method.

  • Measure the gap

    How far is the next threshold?

    Reproduce the rating and show the sources behind it.

  • Check the timing

    When could improvement count?

    Connect the measurement period with the publication and rating year.

Evaluation

Tested on known answers. Before anyone relies on it.

Known cases and the questions reviewers use every day become the test set. Each reviewed answer becomes a repeatable test.

Fig. 6 · Question set by versionIllustrative data
Evaluation results by question and version
Questionv1v2v3
Why did measure M-12 drop from four to three stars?Movement~✓✓
Did the cut point or the rate change?Rule✗✓✓
Which methodology release applies to rating year 3?Rule✗✗✓
Reproduce the rating from the published cut pointsGap✓✓✓
How far is M-12 from the four-star threshold?Gap✓~✓
Which measurement period feeds next year’s rating?Timing✗~✓
Which measures share this data source?Dependencies~✓✓
v3: 7 of 7 pass · release needs all 7
Each row is a question agreed with the team; each column a version of the workflow. The candidate ships only if it holds the release bar.
  • 01

    A reviewer checks the answer

    Correct, partly right or wrong.

  • 02

    It becomes a repeatable test

    Reviewed answers join the set.

  • 03

    Every update runs the same test

    Prompts, tools and sources included.

  • 04

    Only approved versions go live

    A named owner signs off.

Outcomes

Answers with their sources, before the review meeting.

To an explained movement
1 h
from days of file hunting
Gap to next threshold
4 pts
reproduced from cut points
Values source-linked
100%
file, page and release year
Eval pass, v3
7 of 7
release bar
Illustrative data
Deployment

Run it where your data and controls require.

The same evidence trail, agent controls and expert review across four deployment patterns.

We run the application environment for you.

Managed

A dedicated, managed environment with your access controls and audit.

  • Dedicated tenancy
  • Your SSO and roles
  • Full trace and audit

Run in your own cloud tenancy.

Private cloud or VPC

Agent Hive runs inside your private network, connected to your systems through approved paths.

  • Private tenancy and network rules
  • Model access set by agent role
  • One place for traces and evaluation

Keep platform and models in your data centre.

On-premises

The platform, the models and every trace stay on infrastructure you operate.

  • Customer-controlled boundary
  • Approved read and write connectors
  • Evidence and history kept locally

Keep sensitive evidence inside, control model access.

Hybrid model access

A policy gateway controls what is sent, which model receives it and what comes back into the workflow.

  • Model choice per agent role
  • One controlled egress path
  • Graph and sources stay local
Fig. 7 · Deployment options
Deployment options compared: Managed, Dedicated cloud, On-prem, Air-gapped
Runs inQpiAI managed environmentA dedicated network in your cloudYour data centre (Docker / Compose)An isolated network
Operated byQpiAIAgreed per engagementYour teamYour team
TenancyMulti-tenant, isolatedSingle-tenantSingle-tenantSingle-tenant
Data at restManaged environmentYour cloud and regionYour networkInside the perimeter
ModelsManaged modelsApproved endpointsLocal, or via a policy gatewayLocal models only
OutboundOperated by QpiAIPrivate connectionsOne controlled egress path (hybrid)None
IdentityYour identity rulesEnterprise identitySSO and RBACLocal SSO and RBAC
UpdatesRolled out by QpiAIAgreed per engagementNew image version, pulledSigned offline packages
AuditAudit historyAudit historyLocal audit historyLocal audit history

Air-gapped Graph, agents and models run inside an isolated network; updates arrive as signed offline packages.

What runs where in each option. Select a column to move the boundary.

Run this blueprint on your data. We set up the first one with you.