One trusted view of
research sites and investigators
How Inato built one unified, trusted view of research sites and investigators — at a scale a custom-built matching engine could not reach.
Clinical trial technology / healthtech
Entity resolution for research site and healthcare practitioner profiles
Managed Tilores cloud, accessed via the GraphQL API
In production
Connecting trials with the sites that reach patients
Inato is a clinical trial technology platform that connects pharmaceutical sponsors with more than 6,000 research sites — so that trials reach patients in their own communities. Sites need a complete profile to compete for the right trials; sponsors need one to choose where to run a trial with confidence.
That depends on complete, trusted profiles of research sites and the investigators within them. Entity resolution — recognising that records from different sources refer to the same real-world site or practitioner — is foundational to everything Inato's platform does.
“During our trial period we were really successful in autonomously implementing nine distinct entity resolution use cases in three months, which was a faster pace than before.”
Fragmented data, and a matching engine that couldn't scale
The data behind those profiles is fragmented across self-reported, third-party and public-domain sources, each with its own identifiers, formats and levels of completeness. Pulling it into one trusted view is a hard entity-resolution problem.
Inato had built its own matching engine, which served its first use cases for several years. But every new data source took around three weeks to onboard, and the approach could not scale beyond the research sites already held in Inato's master data.
A missing site is a missed opportunity to propose it to a sponsor, so coverage carries a direct commercial impact.
Facing a build-versus-buy decision, the team recognised entity resolution as a specific problem they could outsource — rather than continue to maintain a custom engine that took weeks to onboard each new source and could not reach beyond the data already in hand.
A real proof of concept on their own data
Evaluated against an enterprise master data management solution, Tilores stood out for letting Inato run a genuine proof of concept on its own data, and for being purpose-built for entity resolution rather than feature-heavy. The team's must-haves were clear:
- ▹ A proof of concept on real data with measurable precision and recall
- ▹ Autonomy to maintain and tune the solution without professional services
- ▹ A managed service that works at scale
On every count, a proof of concept with measurable precision and recall — on real data, run by Inato's own team — confirmed the choice.
“Tilores is a managed tool for entity resolution. You ingest records from any source, you define rules, and you get back the records deduplicated. It gives you a lot of control over the matching rules and a lot of explainability over the results, and lets you work at a scale that would be hard to manage internally.”
A fully managed service
Tilores runs as a fully managed service. Data is prepared in Inato's warehouse and fed into Tilores, which performs all of the entity resolution on the backend; Inato delegates the infrastructure and scalability entirely.
The resulting golden records are returned to the warehouse as a cross-source mapping table that powers profile enrichment across the platform — while the matching rules stay in Inato's hands to configure and iterate.
“Tilores makes all the entity resolution on the backend. It is completely managed — we do not need to worry.”
Autonomy and explainability
Because Tilores is configurable rather than a black box, and because the team is autonomous to make changes themselves, Inato has a much faster iteration loop for improving precision and matching quality. The team can tune the matching, see how matches are made, and explain why results changed over time — without waiting on professional services.
Tilores also proved responsive. During the trial it delivered two requested features within weeks — an assembly-time endpoint that signals programmatically when resolution is complete, and version history for matching rules — both now in constant use. Inato points to rapid support over Slack as a further reason for its confidence in the platform.
“We have a faster iteration loop when we want to improve precision and matching quality, because we have a tool that is less of a black box, configurable, and on which the team is autonomous to make changes.”
More reliable data, at a scale they couldn't reach before
Tilores changed both the pace and the scale at which Inato can work, and improved the quality of the data behind its profiles. Inato now reconciles up to around 900,000 research site records and 600,000 healthcare practitioner records together — a scale it could not reach before.
Precision improved on every use case migrated from the legacy engine, and coverage rose by 12 percentage points for the national provider identifier and by 7 percentage points for verified trial experience — meaning more reliable information about more practitioners on the platform. New use cases now stand up roughly 2.5x faster than before.
“Like other modular, managed components of a modern data stack, Tilores allows for rapid time to value and a lower cost of ownership in the long term.”
Build trusted profiles from
fragmented data
See how Tilores resolves records into golden records at scale. Available on AWS Marketplace.
Reconciling data across sources? See the Master Data Management solution.