Top 5 Open-Source MPI Solutions for US FHIR Stacks

Top 5 Open-Source MPI Solutions for US FHIR Stacks

Open-source MPI options for US FHIR stacks have matured to the point where serious health systems and digital health vendors run them in production. The trade-off is the same as for any open-source healthcare component: the licensing fee disappears and the operational load shows up. For US teams that can carry the operational side, these five options are the ones worth a real evaluation in 2026.

The cornerstone Complete US FHIR Master Patient Index Buyer's Guide for 2026 sets the broader frame. For more FHIR coverage for hospital IT, the rest of the desk follows on.

OpenEMPI

OpenEMPI is one of the longest-running open-source MPI projects, with a Fellegi-Sunter probabilistic engine and a history of production use in US health systems. The project has slowed in recent years, but the codebase is still credible for teams that want to fork and own it. For US health systems with a strong identity engineering bench, OpenEMPI is a reasonable starting point.

Mirth Match

Mirth Match, from NextGen Healthcare's open-source lineage, brings a packaged probabilistic matching engine that integrates with the broader Mirth ecosystem. For US health systems already running Mirth Connect for interoperability, Mirth Match is the path of least friction into an open-source MPI.

HAPI Patient Match Extensions

HAPI FHIR's patient matching capabilities have grown over the past two years. The built-in match operation, paired with a deterministic backbone and one of the open-source probabilistic libraries, gives US health systems running HAPI a workable in-platform MPI. The capability is not as deep as a dedicated MPI product, but for FHIR-first US health systems, the integration is tight.

Custom Engine on a FHIR Server

A pattern that has matured for US digital health vendors is a custom matching engine on top of a FHIR server. The team owns the matching logic, the FHIR layer handles the rest, and the resulting product is a FHIR-native MPI tailored to the team's patient population. The work is real, but for US health systems with specific matching requirements, the route gives the most control.

Synthea-Trained Open Source Matching

A handful of US digital health teams have built open-source matching engines trained on Synthea-derived synthetic patient data, then tuned the engine on internal patient data without exposing it to PHI during the model development phase. The pattern fits US digital health startups that need a credible matching engine without the licensing cost of a commercial product and without an early PHI exposure surface.

Picking an Open-Source MPI for a US FHIR Stack

The shortlist for an open-source MPI usually comes down to two questions. Does the team want a packaged engine to fork and run? Then OpenEMPI or Mirth Match. Does the team want an in-FHIR-platform capability? Then HAPI patient match extensions, optionally with a custom engine layer. Does the team have a specific patient population that benefits from tuned matching? Then a custom engine on a FHIR server is the right route, possibly bootstrapped against Synthea synthetic data.

The Commercial vs Open-Source MPI for US ACO deployments covers the trade between open-source and commercial in more detail. The right open-source MPI for a US FHIR stack is the one the team can keep tuned and operational for several years without the licensing cost of a commercial product turning into the engineering cost of an unfinished one.

Sources