Skip to main content

EMPI Lite Overview

Tuva has two EMPI solutions:

  • EMPI Lite: the current primary offering, sold through the Tuva Marketplace
  • Legacy EMPI: the earlier open source implementation built on Splink with a frontend layer

Most teams should start with EMPI Lite. It is faster to implement and easier to maintain. A typical setup takes days, while legacy enterprise EMPI projects can take months.

EMPI Lite is a paid product. We still publish the documentation here so implementation details and behavior are fully transparent.

Legacy EMPI resources:

What EMPI Lite does​

EMPI Lite links patient records across source systems and assigns one persistent empi_id per real person. It runs as dbt SQL in your warehouse.

Core capabilities:

  • Probabilistic matching across demographic attributes
  • Configurable blocking, weights, thresholds, and penalties
  • Connected component clustering for transitive matching
  • Golden record generation for one resolved patient profile per empi_id
  • Human review queues for borderline matches and split candidates
  • Full event history for audit and traceability

What you get​

EMPI Lite is delivered as source code, not a hosted API. You run it in your own dbt environment.

Outputs include:

  • empi_crosswalk for source id to empi_id mapping
  • empi_golden_record for resolved demographics
  • empi_patient_events for explainable event level history
  • Review queues and supporting anomaly tables

Tuva integration​

EMPI Lite produces Tuva compatible patient and eligibility outputs where person_id = empi_id. Point your Tuva input layer to those outputs and downstream marts inherit the resolved identity.

Start here​