Harmonization of various Common Data Models and Open Standards for Evidence Generation Mitra Rocca, Dipl.-Inform. Med. Office of Translational Sciences Center for Drug Evaluation and Research Food and Drug Administration March 14, 2018

1

Agenda       

Project Team Project Goal Problems to Solve Current and Proposed States Project Objectives Modeling Activities Standards Development Activities  Health Level Seven (HL7) Fast Healthcare Interoperability Resources (FHIR)  CDISC SDTM

2

Project Team

3

PROJECT GOAL AND OBJECTIVES Goal: Build a data infrastructure for conducting patient-centered outcomes research using Real World Data (RWD)/observational data* derived from the delivery of health care in routine clinical settings. Objective: Develop the method to harmonize the Common Data Models of various networks, allowing researchers to simply ask research questions on much larger amounts of Real World Data than currently possible, leveraging open standards and controlled terminologies to advance Patient-Centered Outcomes Research. *Observational

Data: Examples of observational data are administrative claims, Electronic Health Records (EHRs), registry, mobile health, electronic patient reported outcomes (ePRO), …).

4

PROBLEMS TO SOLVE Networks of observational data use different CDMs. Open, consensus-based standards might not be leveraged in these CDMs (ex: CDISC, HL7) There is a need to facilitate interoperability among these collaborations

5

Current State: Limited groups of connected Real World Data systems  Different networks use various sources of healthcare and research data.  Creation of different mini-networks for research  Allows researcher to query (ask question) and simultaneously, receive results from many different sources. Q Sentinel

BUT: Each network communicates with different agreed upon methods: “Common Data Model”

A

Data Partners

6

The solution, using the Adapter Analogy Sentinel

I2b2/ACT

PCORNET

OMOP

 Different countries use different “outlets”.  There is a need for travel adapters. The Solution:  Use a converter between various adapters.  Allow researchers to ask a question once and receive results from many different sources using a common agreed-upon standard structure, or a Common Data Model.7

Proposed Solution

8

8

Modeling Activities for the Common Data Model Harmonization (CDMH) Project

9

Biomedical Research Integrated Domain Group (BRIDG) Implementation Approaches  Reference Model (ISO, HL7, CDISC Standard)  Source for clinical research data semantics & foundation model  Data Integration/Mapping Solutions  One mapping to a standard rather than multiple point to point mappings  Exchange Format  Subsets of BRIDG classes represented in XSD/XML  Physical Database  Develop BRIDG-based logical and physical database model 10

Leveraging BRIDG model on CDMH Data Integration/Mapping Solutions:  BRIDG mapped to CDISC SDTM and HL7 FHIR (STU3)  Mapped 5 networks Common Data Models to BRIDG • PCORnet (v3.1 and v4.0), Sentinel, i2b2, OMOP

11

BRIDG Model in CDMH – Wheel & Spoke

12

BRIDG-based CDMH Model to HL7 FHIR  BRIDG 5.0.1 is being mapped to HL7 FHIR (STU3)  FDA Grant Award (U01FD005846)  CDMH project leveraging the Grant work to get the initial set of BRIDG-based CDMH model to HL7 FHIR mappings  Beginning the task of identifying the gaps to document the extensions and profiles needed on existing FHIR resources

13

BRIDG to HL7 FHIR (STU3) Mapping (leveraging FDA Grant work - U01FD005846)

14

CDMH DRAFT list of HL7 FHIR Resources • • • • • • • • • •

Person Patient Research Subject Organization Location Practioner Practioner Role Activity Definition Plan Definition Task

• • • • • • • • • • •

Device Drug Observation Condition Procedure Medication Specimen Medication Dispense Medication Administration Adverse Event Immunization 15

Mapping Levels and Model Development Steps Physical Models

1

Conceptual Model

Mappings

2 CDMH BRIDG View

3 CDMH Logical Model

Physical Models

5 Bidirectional Mappings

Conceptual Model

(Subset of BRIDG Classes PLUS New semantics from 4 models)

BRIDG-Based Logical Model

4 CDMH Repository

BRIDG-Based Physical Model 16

Mapping Validation Process Steps Mapped the Common Data Models (CDMs) to BRIDG 5.0.1  Sentinel - DONE  Patient-Centered Outcomes Research Network (PCORNET) v3 – DONE  Patient-Centered Outcomes Research Network  (PCORNET) v4 - DONE  Observational Medical Outcomes Partnership (OMOP) – DONE  Informatics for Integrating Biology & the Bedside (i2B2) / Accrual for Clinical Trials (ACT)– Planned 17

Mapping Path Validation

18

DRAFT CDMH BRIDG View Clinical Trial related

DRAFT / WORK-IN-PROGRESS CDMH Conceptual BRIDG View

19

Benefits to Public Health and Research  Evaluates potential safety concerns  Supports provisions of 21st Century Cures Act  Supports the Patient Centered Outcomes Research goals  Enhances regulatory decisions by providing FDA reviewers with access to a larger network of RWD data  Supports the goals outlined in the All of us (Precision Medicine Initiative) 20

Thank You

21

Back up

22

Real World Data: Definition

Claims Data

Laboratory Results

Electronic Health Records

Real World Data

Registries

Vital Records

Mobile Devices

RWD include data derived from electronic health records (EHRs), claims and billing data, data from product and disease registries, patient-generated data including in home-use settings, and data gathered from other sources that can inform on health status, such as mobile devices. 23

Real World Evidence: Definition RWE is the clinical evidence regarding the usage and potential benefits or risks of a medical product derived from analysis of RWD.

24

RWE: What are the Goals? Maximize the opportunities to have regulatory decisions incorporate data/evidence from settings that more closely reflect clinical practice  Increase the diversity of populations  Improve efficiencies • Population identification/selection • Reduce duplicative capture of data

25 25