Deep dive — methodology
Counterparty-question rehearsal — the 5-decision reproducibility drill
Every $499 Snapshot samples five AI-driven decisions from a target-day and reproduces them defensibly. Not because you need them all today — because the discipline of reproducing five proves the discipline could reproduce any of them tomorrow.
The prompt the rehearsal answers
"Reproduce this AI-generated triage decision from Day 47 including the patient's chief complaint category, admission source, comorbidity profile, model version at decision-time, triage score, and lane routed — as a defensible record we can produce for HHS OCR investigation, CMS survey response, malpractice defense counsel, or State AG AI-oversight review."
The counterparty question — identical shape whether it comes from OCR, plaintiff, surveyor, or carrier
Sample encounter decisions from Day 47
| Encounter ID | Score | Lane | Model version | Decision hash |
|---|
| E-003040 | 77.69 | Urgent Intervention | ed-triage-classifier-v3.1.2 | edt-003040 |
| E-003041 | 64.63 | Standard Workup | ed-triage-classifier-v3.1.2 | edt-003041 |
| E-003042 | 45.65 | Standard Workup | ed-triage-classifier-v3.1.2 | edt-003042 |
| E-003043 | 65.41 | Standard Workup | ed-triage-classifier-v3.1.2 | edt-003043 |
| E-003044 | 47.64 | Standard Workup | ed-triage-classifier-v3.1.2 | edt-003044 |
What "defensibly reproduced" means
- Model version pinned. Not "the current model" — the exact device-software-function version deployed at decision-time.
- Input snapshot bound. Chief complaint category + admission source + comorbidity profile + LOS estimate exactly as they were when the AI scored the encounter.
- Decision hash bound. Cryptographic hash tying the input to the output — makes tampering detectable.
- Retention pipeline independent. The retention system that holds the decision records is not the production model itself — independence of the record from the actor being recorded.
What the rehearsal proves
All 5 sampled Day-47 decisions reproduced with defensible-records match. Model version pinned, decision hash cryptographically bound, input data retained via decision-provenance record. If OCR / plaintiff / surveyor / carrier requests any of the ~5,996 encounters in the audit period, the same reproduction procedure applies.
Retention horizon
- HIPAA 45 CFR 164.316: 6-year minimum on documentation of policies + records
- State-specific malpractice tail: typically 7-10 years for clinical AI
- HHS OCR Section 1557 investigation-hold: discipline of preserving through resolution
- Litigation hold: preserves through matter conclusion, may extend original horizon
The rehearsal record is not a one-time exhibit — it is proof that the retention discipline is operational.
Why 5, not 500
Reproducing 500 decisions on request is a discovery-response exercise, appropriately triggered by a specific legal process. Reproducing 5 sample decisions at Snapshot time is a discipline-test: it validates that the retention pipeline works, without exhausting the audit budget on record production nobody has asked for yet.
$499 Snapshot. 3 business days.
Same 5-decision reproducibility drill on your hospital's actual AI surface + counterparty-question rehearsal record.
Buy $499