Cross-reference hospital MRF median allowed amounts against payer TiC in-network rates to identify charge-rate discrepancies for price transparency analytics
Ingest hospital MRF data (charge type, billing code, payer name, plan name, median allowed amount, 10th and 90th percentile allowed amounts) and payer TiC in-network rates (billing code, negotiated rate, payer name, plan name) into a common data store.
Match records on (billing_code, billing_code_type, payer_name, plan_name) using fuzzy matching on name fields to account for abbreviation and spelling differences between the two file sources.
For hospital records where the charge type is percentage-based, use the median allowed amount as the dollar comparator rather than the percentage figure; for flat-rate entries, use the negotiated rate directly.
Calculate the percentage difference between the hospital-reported median allowed amount and the payer-reported negotiated rate for each matched pair; flag pairs where the difference exceeds a configurable threshold.
Group outliers by payer and service line to identify systemic discrepancies that may indicate data quality issues, stale MRF data, or genuine rate disagreements between payer and provider filings.
Report findings with both the hospital MRF last_updated date and the payer MRF last_updated date so that temporal misalignment can be distinguished from substantive discrepancies.
Known gotchas
Hospital MRFs and payer MRFs are produced independently on different schedules; a hospital may post an April MRF while the matching payer file still reflects January rates, making apparent discrepancies date-driven rather than genuine.
The same CPT or DRG code may appear in both files with different rate values if the hospital and payer use different billing code versions (for example, ICD-10-PCS versus CPT); match on code type as well as code value.
Payer in-network files may express rates as fee schedule percentages rather than dollar amounts; converting these to dollars requires the applicable fee schedule base (for example, Medicare fee schedule values), which is a separate data dependency.
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp