The API Layer That Crosses Silos Without Moving Data

by Marina VasquezHead of Semantic Systems, HMND

This week’s movement around CDISC APIs, HL7 FHIR, and real-world data (RWD) interoperability isn’t about new platforms; it’s about entity resolution across clinical, preclinical, and manufacturing systems, graph-native mapping of trial outcomes to targets, and APIs that span silos without moving data . The concrete semantic layer is a federated ontology that binds Study, Site, Participant, and Target across domains using FHIR resources as the transport and CDISC models (SDTM, ADaM) as the content semantics . Trial teams still join data manually, and that breaks at scale—months of endpoint delay are the cost of a trial where silos force human joins . The IT friction here is real: health systems often lack the schema alignment or API contracts to let FHIR and CDISC talk without a custom script written by a tired data engineer.

Why Adoption Is Hard: Governance and Tight Contracts

Adoption stalls because semantic governance requires cross-functional ontology ownership. No single team can own the ontology that spans clinical, preclinical, and manufacturing without a council that includes pharmacologists, statisticians, QA, and data engineers. API contracts must be tight across domains: a FHIR ResearchStudy resource must align with CDISC’s SDTM study metadata, and a manufacturing Batch must resolve to the same trial participant without ambiguity . When contracts are loose, systems drift, and joins become manual again. In pharma, this means IT can’t just spin up a microservice; they need a regulated ontology council that signs off on every field mapping, because an FDA audit will ask for the trace from raw data to submission. That review cycle alone kills velocity.

Failure in the Real: A Trial That Waits Months

Failure looks like a trial where data silos force manual joins. The EDC system holds participant visits; the CTMS holds site timelines; the eTMF holds IRB approvals; the lab system holds preclinical biomarkers; the manufacturing system holds batch-release data. Without a shared semantic layer, teams must write scripts to join these sources, validate mappings, and re-join when definitions shift. Endpoints are delayed, and months are wasted—precisely the inefficiency CRISI aims to eliminate by standardizing operational data exchange with FHIR . The failure isn’t just slow; it’s brittle. One definition change in the lab system can break the entire join chain, and the team spends weeks debugging Python scripts instead of analyzing biology.

What Works: Graph-Native Mapping and Federated APIs

The stack that works uses knowledge graphs to model relationships between entities and rules, and open APIs to query standards dynamically without moving data . CDISC validation rules stored as graph structures let AI reason over relationships (e.g., “this outcome maps to this target, which maps to this batch”) and execute rule validation in a graph database . RESTful APIs return JSON so clinical systems can pull metadata, rules, and standards in real time, avoiding manual import . The result is one question across silos: “Show me all outcomes for participants exposed to this target in this batch,” answered by the API layer, not by a platform that aggregates data. For IT, this means fewer ETL pipelines, fewer data lake copies, and a query interface that actually respects the regulatory chain of custody.

The Bridge Is Not a Platform

The bridge isn’t a platform; it’s the API layer that lets you ask one question across silos. It’s the federated contract where FHIR resources carry the payload, CDISC models define the semantics, and the graph resolves the ontology. When that layer is tight, manual joins disappear, endpoints accelerate, and drug development becomes efficient again . If you’re staring at the same manual join problem in your next trial, maybe the bridge is just the API contract you haven’t signed yet.