Audit Exposure & Extrapolation Engine
The 2026.04 Data Version is a directional diagnostic model calibrated for New York State Social Care Networks (SCNs) under the NYS 1115 Waiver framework. It quantifies systemic documentation leakage and audit exposure using deterministic, reproducible formulas with fixed constants and no stochastic generation.
Verification note: This page provides formula-level transparency to address black-box concerns for CBO operators, SCN leads, and auditors. All calculations are reproducible from the constants documented below.
This model is calibrated for Medicaid-aligned Social Care Network (SCN) reconciliation flows under the NYS 1115 Waiver. It is not validated for Medicare Advantage (MA) risk adjustment, encounter data submission, or RADV audit mechanics. MA claim workflows, coding hierarchies, and audit methodologies differ substantially and require distinct analytical frameworks.
Network node performance is modeled using deterministic index-driven distribution across the documented ranges (screening: 21.6%–95.9%; referral completion: 86%–96%). These ranges are anchored to aggregated SCN performance data. Actual node performance will vary by geography, capacity, and operational maturity.
All outputs are directional estimates, not predictions of actual financial outcomes or audit liability. The model identifies systemic risk patterns and proportional exposure — it does not constitute legal, audit, or compliance advice. Confidence bands (±15%) reflect modeled uncertainty, not actuarial ranges.
| Parameter | Value | Source / Basis | Override? |
|---|---|---|---|
| V (monthly encounter volume) | User input (1–100,000) | Operator-specified | Yes — dashboard slider |
| $9.32 (ENCOUNTER_COST) | $9.32 per encounter | Validated 18-minute documentation burden baseline | No — fixed constant |
| 0.28 (MEAT_FAILURE) | 28% | MEAT failure rate from 50,000-submission audit reference pattern | No — fixed constant |
| 12 (annualisation) | 12 months | Converts monthly to annual impact | No — structural |
Interpretation: Lop represents annualised value leakage caused by non-compliant documentation fidelity under the MEAT framework (Monitoring, Evaluation, Assessment, Treatment).
| Parameter | Value | Source / Basis | Override? |
|---|---|---|---|
| 0.12 (DENIAL_RATE) | 12% | National benchmark denial probability for Medicaid-like claim flows | No — fixed constant |
| $57.23 (REWORK_COST) | $57.23 per denied event | Validated administrative burden per denied encounter requiring rework | No — fixed constant |
Interpretation: Ldenial captures direct rework losses from denials and downstream administrative reprocessing. Denial patterns in this model are treated as systemic, not isolated events.
| Parameter | Value | Source / Basis | Override? |
|---|---|---|---|
| 0.885 (AUDIT_FAILURE_RATE) | 88.5% — mid-point of 81%–96% OMIG targeted-review range | Observed in targeted high-risk audit populations under OMIG-style review | No — fixed constant |
| scale (non-linear volume amplifier) | 1 + log₁₀(V/1000) when V > 1000, else 1 | Captures audit exposure amplification at high encounter volumes | No — derived from V |
| 0.10 (AUDIT_PROBABILITY) | 10% | Modeled annual probability for high-risk profile audit targeting | Audit page (custom runs only) |
Interpretation: Eaudit estimates statistically extrapolated annual risk exposure under targeted audit mechanics. The 0.885 failure rate represents the mid-point of the 81%–96% OMIG-observed range for targeted high-risk audit populations. A non-linear scale factor amplifies exposure for volumes exceeding 1,000 encounters/month. The operational risk threshold for node remediation routing is fixed at 0.58.
Sfid = (Screening × 0.4) + (Referral × 0.3) + (0.72 × 0.3)
| Component | Weight | Data Range | Basis |
|---|---|---|---|
| Screening Completion Rate | 40% | 21.6% – 95.9% | Aggregated SCN performance anchors |
| Referral Completion Rate | 30% | 86% – 96% | Aggregated SCN performance anchors |
| Documentation Integrity (1 − MEAT_FAILURE) | 30% | Fixed: 0.72 | Derived from MEAT failure rate (1 − 0.28 = 0.72) |
Node distribution is deterministic and index-driven. No random() is used. The high-risk threshold (0.58) triggers audit exposure classification for any node scoring below that boundary.
Block Hash Formula
hash = SHA256(prev_hash + JSON(payload) + timestamp)
Execution records are chained with SHA-256 (Web Crypto API, no external libraries). Each block stores: index, timestamp, payload, payload hash, previous hash, and current hash. This produces append-only lineage semantics compatible with WORM evidentiary posture.
Chain verification recomputes every block hash end-to-end. If any hash does not match its recomputed value, the system displays "System Integrity Compromised". Client-side chain simulation is provided for transparency; the canonical immutable record is persisted server-side in append-only JSONL form.
Low = value × 0.85 | High = value × 1.15
All outputs display a ±15% confidence band to reflect modeled uncertainty in operational benchmarks and audit targeting rates. These bands are directional and do not represent actuarial confidence intervals or statistical significance bounds.
The model enforces explainability, reproducibility, and internal monitoring support so compliance officers can demonstrate structured oversight, control testing, and corrective action triggers tied to measurable documentation risk.
The methodology is designed for minimum-necessary data exposure, auditable access boundaries, and privacy-preserving handling of sensitive social and behavioural health context in both simulation and persistent ledger implementations.
Deterministic performance weighting and leakage calculations align with SCN accountability expectations by translating documentation fidelity into explicit network remediation logic and audit defensibility metrics.