Worked example: a GOOD carrier
Status: demonstrated — a real carrier from the prototype database.
This is a complete GOOD arc exactly as it exists in the database of the 2026 synthetic proof of concept: carrier 21752 from pool 5, a 34-year-old woman whose 48-row year of claims hides four scattered SIGNALs, one TEST, and a quiet run of refills afterward.
The year at a glance
Read without annotations, this is an unremarkable year of care. Its backbone is chronic low back pain: general-practice visits, a long run of chiropractic sessions, ibuprofen and later naproxen refills, a lumbar X-ray, eventually an MRI. Around it sits the ordinary texture of primary care — telehealth visits that pick up a depression code, a lipid panel after a hypercholesterolemia mention, an A1C for type 2 diabetes, an annual preventive exam with a screening mammogram. None of this has anything to do with Primary Veladrin Excess; it is the synthetic base history the phenotype was injected over.
The injection adds seven rows. Four are SIGNALs, spread across five months: a 24-hour urine mineral-excretion panel on the very first row, a gynecologic mineral-axis evaluation six days later, a February pharmacy fill of hydralozide — one of PVE's fabricated antihypertensives, coded to resistant blood pressure — and extended ambulatory blood-pressure monitoring at a cardiology practice at the end of May. Then, at row 27 of 48, the TEST: CPT-82197, the serum veladrin ratio, drawn at an independent laboratory. After that, the entire remaining story of the condition is two 90-day refills of veladrostat, in July and October. That is what a GOOD arc is: signals, a test in time, and a condition managed so well it never accrues an IMPACT or LATE row. (Contrast the BAD carrier, where the test comes late and the timeline ends in recurring facility care.)
The timeline, annotated
The leading annotation column is construction bookkeeping: 🟡 SIGNAL · 🟠 IMPACT · 🔵 TEST · 🟢 EARLY · 🔴 LATE, empty for background rows. No model ever sees these marks — or any pool or patient identifier; models see exactly the six columns and nothing else. A GOOD carrier has no 🟠 or 🔴 rows by definition. … rows stand for skipped stretches, and some long WHAT descriptions are truncated for width.
| PERSON | WHEN | WHERE | WHO | WHAT | WHY | |
|---|---|---|---|---|---|---|
| 🟡 | 34 female | 2025-01-06 | POS-81: Independent Laboratory | NUCC-291U00000X: Clinical Medical Laboratory | CPT-83541: 24-hour urine mineral-excretion panel | ICD-E87.A3: Recurrent electrolyte mineral derangement; |
| 34 female | 2025-01-06 | POS-11: Office | NUCC-208D00000X: General Practice | CPT-99213: Office or other outpatient visit… | ICD-M54.50: Low back pain, unspecified; | |
| 34 female | 2025-01-10 | POS-11: Office | NUCC-111N00000X: Chiropractor | CPT-98941: Chiropractic manipulative treatment (CMT); spinal, 3-4 regions | ICD-M54.50: Low back pain, unspecified; | |
| 🟡 | 34 female | 2025-01-12 | POS-11: Office | NUCC-207V00000X: Obstetrics & Gynecology | CPT-83543: Gynecologic mineral-axis evaluation | ICD-N94.A1: Female pelvic mineral-axis symptom; |
| … | … | … | … | … | … | … |
| 34 female | 2025-02-10 | POS-11: Office | NUCC-208D00000X: General Practice | CPT-99213: Office or other outpatient visit… | ICD-J06.9: Acute upper respiratory infection, unspecified; ICD-R05: Cough; | |
| 🟡 | 34 female | 2025-02-15 | POS-01: Pharmacy | NUCC-333600000X: Pharmacy | ATC-C03AZ08-211584: hydralozide 25 MG Oral Tablet [#30 DS:30] | ICD-R03.A1: Resistant blood pressure elevation on multiple agents; |
| 34 female | 2025-02-17 | POS-11: Office | NUCC-111N00000X: Chiropractor | CPT-98940: Chiropractic manipulative treatment (CMT); spinal, 1-2 regions | ICD-M54.50: Low back pain, unspecified; | |
| … | … | … | … | … | … | … |
| 34 female | 2025-05-12 | POS-11: Office | NUCC-261QR0200X: Clinic/Center, Radiology | CPT-72148: Magnetic resonance (eg, proton) imaging, spinal canal and contents, lumbar; without contrast material | ICD-M54.50: Low back pain, unspecified; ICD-M51.36: Other intervertebral disc degeneration, lumbar region; | |
| 34 female | 2025-05-20 | POS-11: Office | NUCC-208D00000X: General Practice | CPT-99213: Office or other outpatient visit… | ICD-M54.50: Low back pain, unspecified; ICD-E78.00: Pure hypercholesterolemia, unspecified; | |
| 34 female | 2025-05-20 | POS-11: Office | NUCC-207Q00000X: Family Medicine | CPT-80061: Lipid panel… | ICD-E78.00: Pure hypercholesterolemia, unspecified; | |
| 🟡 | 34 female | 2025-05-30 | POS-11: Office | NUCC-207RC0000X: Internal Medicine, Cardiovascular Disease | CPT-93785: Extended ambulatory blood-pressure monitoring | ICD-R03.A1: Resistant blood pressure elevation on multiple agents; |
| 34 female | 2025-06-03 | POS-01: Pharmacy | NUCC-333600000X: Pharmacy | ATC-M01AE02-206165: Naproxen 500 MG Oral Tablet [#30 DS:30] | ICD-M54.50: Low back pain, unspecified; | |
| 🔵 | 34 female | 2025-06-06 | POS-81: Independent Laboratory | NUCC-291U00000X: Clinical Medical Laboratory | CPT-82197: Serum veladrin / regulatory-mineral ratio, quantitative | ICD-E27.A4: Suspected mineralocorticoid-axis dysregulation; |
| 34 female | 2025-06-16 | POS-11: Office | NUCC-111N00000X: Chiropractor | CPT-98940: Chiropractic manipulative treatment (CMT); spinal, 1-2 regions | ICD-M54.50: Low back pain, unspecified; | |
| … | … | … | … | … | … | … |
| 🟢 | 34 female | 2025-07-04 | POS-01: Pharmacy | NUCC-333600000X: Pharmacy | ATC-C03DZ01-211582: veladrostat 12.5 MG Oral Tablet [#90 DS:90] | ICD-E27.A5: Primary veladrin excess; |
| 34 female | 2025-07-14 | POS-01: Pharmacy | NUCC-333600000X: Pharmacy | ATC-M01AE02-206165: Naproxen 500 MG Oral Tablet [#30 DS:30] | ICD-M54.50: Low back pain, unspecified; | |
| … | … | … | … | … | … | … |
| 34 female | 2025-08-25 | POS-11: Office | NUCC-208D00000X: General Practice | CPT-99213: Office or other outpatient visit… | ICD-E11.9: Type 2 diabetes mellitus without complications; | |
| 34 female | 2025-08-25 | POS-11: Office | NUCC-207Q00000X: Family Medicine | CPT-83036: Hemoglobin; glycosylated (A1C) | ICD-E11.9: Type 2 diabetes mellitus without complications; | |
| … | … | … | … | … | … | … |
| 34 female | 2025-09-15 | POS-11: Office | NUCC-207Q00000X: Family Medicine | CPT-77063: Screening digital breast tomosynthesis, bilateral… | ICD-Z12.31: Encounter for screening mammogram for malignant neoplasm of breast; | |
| 🟢 | 34 female | 2025-10-02 | POS-01: Pharmacy | NUCC-333600000X: Pharmacy | ATC-C03DZ01-211582: veladrostat 12.5 MG Oral Tablet [#90 DS:90] | ICD-E27.A5: Primary veladrin excess; |
| … | … | … | … | … | … | … |
| 34 female | 2025-12-18 | POS-02: Telehealth Provided Other than in Patient's Home | NUCC-208D00000X: General Practice | CPT-99213-95: Office or other outpatient visit…; Synchronous telemedicine service via real-time interactive audio and video | ICD-M54.50: Low back pain, unspecified; ICD-F32.9: Major depressive disorder, single episode, unspecified; |
What to notice
Four things. First, the SIGNAL rows do not cluster — they are scattered among the back-pain visits and pharmacy refills, weeks to months apart, each one individually easy to read past. Nothing announces them as related; a reader would need to already know what a hydralozide fill and a urine mineral panel have in common. Second, the TEST happens where such a test would actually be billed: an independent laboratory, under a "suspected" diagnosis code — in-world, this is the moment somebody thought to check. Third, everything PVE-related after the TEST is just a pill: veladrostat refills on a 90-day cycle, now coded to the confirmed diagnosis. Early detection leaves almost no billing footprint, and that quietness is the whole point of the EARLY outcome. Fourth, everything else — 41 of the 48 rows — is background, and stays background.
What this carrier trains
For secondary training, this timeline is cut at the TEST. The prompt is everything before it, so the prompt ends in SIGNAL territory: all four 🟡 rows in view, test nowhere in sight. The preferred continuation is the 🔵→🟢 sequence — fire the TEST, then the treatment. The lesson in one sentence: when this lead-up pattern is present, the right next rows are the test and the pill. How prompt, chosen, and rejected material is constructed for all four kinds is owned by three-pattern learning.
This carrier trained pool 5's model and no other. For the other four models it is out of distribution — and that is the only capacity in which it counts as evidence. In the prototype's full run, all 250 GOOD carriers — this one included — fired unanimously, 5 of 5; see the results.
Randomized, not choreographed
The arrangement above was drawn at random, not staged. Which patient became a carrier, where the TEST landed, how the SIGNALs spread out — all randomized within loose construction constraints, precisely so a model can key on nothing but the injected pattern itself (the isolated signal). The gynecologic evaluation six days after the urine panel, or the quiet three and a half months between the hydralozide fill and the ambulatory monitoring, encode no clinical claim — PVE has no mechanism to be faithful to. Any choreography a reader senses is just the arc's structure — signals, then test, then treatment — and nothing more.
See also
- Primary Veladrin Excess — the full code inventory behind the injected rows
- Three-pattern learning — how arcs become training material
- The BAD carrier and the NOPE carrier — the other worked examples
- Prototype results — where the 5-of-5 numbers live