How It Works

From EHR Signal to Enrolled Patient

TrialVyx's 4-phase pipeline turns de-identified EHR data into pre-screened trial referrals — shortening the recruitment cycle from months to weeks.

The 4-Phase Pipeline

How the TrialVyx Engine Works

Each phase is designed to minimize latency and maximize the signal-to-noise ratio in patient identification.

Phase A Data Ingestion FHIR R4 · De-ID Phase B Phenotype Extraction Phase C Protocol Matching Phase D Site Referral Pre-Screened EHR platforms NLP + structured signals I/E criteria mapping Site receives referral

A.Data Ingestion

TrialVyx connects to partner EHR systems via FHIR R4 API. Only de-identified phenotypic data is extracted — no PHI, no patient identifiers leave the originating system. The integration layer handles data normalization across different EHR schema versions.

B.Phenotype Extraction

The phenotype engine processes structured (diagnoses, lab values, medication history) and semi-structured (clinical notes NLP) EHR signals to build an anonymized patient phenotype profile. These profiles are scored against trial eligibility dimensions.

C.Protocol Matching

Trial inclusion/exclusion criteria are mapped to phenotype dimensions. The matching engine scores each phenotype profile against open protocol criteria in real time, updating continuously as protocol amendments are submitted. Scores reflect both hard exclusions and soft eligibility signals.

D.Site Referral

Sites receive structured, pre-screened referrals with match confidence scores and a summary of the key eligibility signals that triggered the referral. Coordinators can review the matching rationale before contacting the patient — reducing screen failure from the first point of contact.

EHR Compatibility

FHIR-Compatible EHR Integration

TrialVyx integrates with major EHR platforms via FHIR R4 API. De-identification happens at the source — no PHI ever enters the TrialVyx platform.

  • Epic FHIR R4 compatible
  • Oracle Health / Cerner integration
  • De-identification pipeline — no PHI transmission
  • Designed with HIPAA privacy principles in mind

Data Privacy Architecture

TrialVyx is designed with data privacy as a first principle. De-identification occurs at the EHR integration layer before any data flows into the matching engine.

TLS encryption in transit No PHI storage HIPAA-principled design Least-privilege access
AI Matching Engine

Protocol-Aware Pattern Matching at Scale

Eligibility criteria are complex, ambiguous, and constantly updated. TrialVyx's matching engine translates clinical protocol language into computable phenotype criteria — and updates match scores in real time as protocols evolve.

  • Inclusion/exclusion criteria parsed from protocol documents
  • Match scores updated continuously as patients' clinical status changes
  • Hard exclusion flagging prevents ineligible referrals
Abstract pattern matching visualization
Platform Results

What the Numbers Show

47%
Screen Failure Rate Reduction
Median across deployments vs. pre-TrialVyx baseline
9 mo
Average Enrollment Window
vs. 18-month industry baseline for comparable indications
3.2x
More Trial-Ready Patients
Per site per quarter compared to traditional referral channels

Ready to Accelerate Enrollment?

Start with a pilot deployment. 90 days. Up to 5 sites. See the EHR phenotyping difference firsthand.

Request Pilot Access