Bitemporal: Definition and Healthcare Context
Full name: Bitemporal Data Modeling
Bitemporal data modeling records two independent time dimensions for every fact: valid time — when the fact was true in the real world — and transaction time — when the fact was recorded in the database. A bitemporal store can answer both "what was true on a past date" and "what did our data say on a past date," even after a record is later corrected. For provider data this means a query can reconstruct exactly how a federal source described a provider as of any prior point, which is the basis for point-in-time and as-of lookups.
How it’s used
- CMS NPPES NPI Registry: NPPES publishes weekly and monthly files, so a provider's recorded address has both a real-world effective date and a capture date Fonteum stores separately.
- An append-only store keeps every prior snapshot, so a corrected value never overwrites the record that was live before the correction.
- Point-in-time API reads accept an as-of date and return the data exactly as it stood then — the foundation of an audit-grade historical lookup.
Frequently asked questions
- What is bitemporal data?
- Bitemporal data tracks two timelines for every fact — when it was true in the real world (valid time) and when it was recorded in the database (transaction time) — so history stays reconstructable after corrections.
- What is the difference between valid time and transaction time?
- Valid time is when a fact actually held in the real world; transaction time is when the system learned and recorded it. They differ whenever data is recorded late or corrected after the fact.
- Why does provider data need bitemporal modeling?
- Federal sources revise records retroactively, so an audit must show what the data said on a past date, not just what it says now. A bitemporal store answers as-of queries against any prior point.
Related terms
Explore in Fonteum
How Fonteum sources, resolves, and publishes data tied to this term.