Clear Rates Methodology

The Clear Rates pipeline turns raw payer and hospital machine-readable files into a single, scored, canonical rate per rate object. This section documents each stage of that pipeline — how rates are collected, scored, selected, and delivered.

Pipeline Stages

Stage	What it does	SQL
Spines	Build canonical reference tables for payers, networks, providers, and codes	source
Rate Object Space	Compute the Cartesian product of valid (payer, provider, code, network) combinations	source
Raw Data	Ingest payer MRF, hospital MRF, claims, and gross charges; map onto the rate object space	source
Transformations	Standardize non-dollar rates (percentages, per diems, drug dosages) into comparable dollar amounts	source
Imputations	Estimate missing rates using a hierarchical fallback chain (RC global → RC HCPCS → carveouts → DRG → CSTM)	source
Accuracy	Score every rate 0–7 using outlier bounds, benchmark comparisons, and payer/hospital counterparty validation	source
Rate Selection	Pick the highest-scoring rate per ROID; translate to canonical score 1–5	source
Output	Merge sub-version outputs into production tables; enrich with whispers and rollup views	source

Key Concepts

Spines — curated reference tables that define the universe of payers, providers, networks, and codes. Spines standardize identifiers across data sources and determine which entities appear in the Rate Object Space. See Spines.

ROID — Rate Object ID. A unique combination of (payer_id, network_id, provider_id, billing_code, billing_code_type, bill_type). Every table in the pipeline is keyed on ROIDs.

Canonical — the final, selected values that represent Clear Rates' best answer for a ROID. canonical_rate is the single dollar rate chosen by the Rate Selection algorithm, and canonical_rate_score (1–5) is its confidence score. See Score Hierarchy for the full mapping.