SMART on FHIR Integration Across Epic, Cerner, and eClinicalWorks

Client is a healthcare analytics ISV specializing in AI-driven provider performance insights and clinical quality benchmarking. Their platform needed to aggregate structured, high-fidelity patient data from multiple EHRs to enable predictive and comparative analytics across organizations.

The Challenge.

Before working with Pegasus One, Client had no direct integration with EHR systems. The team relied heavily on manual data ingestion and non-standardized inputs, which limited scalability and the performance of their AI models. Key challenges included:


  • Manual Work: Data was manually imported via Excel sheets, CDA documents, and PDFs , each EHR provided data in different formats.
  • Model Underperformance due to insufficient data: AI models underperformed because key data elements like lab results, diagnostic reports, and medications were missing or incomplete.
  • Inconsistent standards across EHR: Maintaining a consistent data standard across Epic, Cerner, and eClinicalWorks was impossible due to FHIR version discrepancies.
  • No Multi-Tenant: The same app needed to work for multiple EHR systems (Epic, Cerner, and eCW), demanding a multi-tenant architecture capable of dynamic configuration and isolation.

Solution.

SMART on FHIR Integration Application 


Pegasus One designed and implemented a SMART on FHIR-based backend architecture enabling Client to ingest and normalize clinical data across all three EHRs. 

 

Key Highlights.

Multi-EHR, Multi-Tenant Integration

  • One unified backend app supporting Epic, Cerner, and eClinicalWorks (eCW).
  • Backend bulk API integration for EPIC, Cerner & eCW.

FHIR Data Normalization

  • Mapped across FHIR R3 and R4 standards, addressing version inconsistencies between EHR vendors.
  • Standardized NDJSON & JSON schema for downstream analytics.

Data Pipeline and Storage

  • Built an ingestion pipeline for FHIR-to-warehouse data transformation and normalization.
  • Implemented a de-identified AI-ready layer for secure model training and inferencing.

PDF & Clinical Note Processing

  • Imported DocumentReference PDFs and clinical notes into an NLP pipeline to extract insights used in model predictions.

18 Core FHIR Resources Integrated

  • Patient, Encounter, Observation, Condition, Procedure, MedicationRequest, Practitioner, PractitionerRole, DocumentReference, DiagnosticReport, Immunization, AllergyIntolerance, Organization, Appointment, Slot, Location, ImagingStudy and CarePlan.

Key Highlights.

Metric Before After
Average Query Response Time Manual (Excel/CDA) Automated via FHIR APIs
Research Query Success Rate Per EHR Unified across R3/R4
Reduction in Manual Data Review ~60% >95%
Analyst Productivity Gain ~70% ↑ 90% (Post FHIR pipeline)

Conclusion.

The SMART on FHIR integration established a robust, scalable, and AI-ready data ecosystem for Client. It replaced fragmented manual imports with a real-time, standards-based pipeline that fuels predictive analytics and provider benchmarking across multiple EHR environments.

What We Do

Solutions For

Resources

© 2026 Pegasus One Health. All rights reserved.