32
ll Rights Reserved, Duke Medicine 2007 Duke Cancer Institute Involvement with Data Standards and Semantics (May 6, 2011) Salvatore Mungal Duke Bioinformatics Shared Resources (DBSR) Duke Cancer Institute (DCI)

All Rights Reserved, Duke Medicine 2007 Duke Cancer Institute Involvement with Data Standards and Semantics (May 6, 2011) Salvatore Mungal Duke Bioinformatics

Embed Size (px)

Citation preview

All Rights Reserved, Duke Medicine 2007

Duke Cancer InstituteInvolvement with

Data Standards and Semantics(May 6, 2011)

Salvatore MungalDuke Bioinformatics Shared Resources (DBSR)Duke Cancer Institute (DCI)

All Rights Reserved, Duke Medicine 2007

Agenda• Background:

– The Promise of Interoperability– The caBIG Initiative– Interoperability Fundamentals

• Groups of Standards• Standards at Duke• caBIG Interoperability Development• Benefits of caBIG Interoperability Adoption• Building Blocks/Development• caBIG Implementation Framework at Duke• Benefits of caBIG Adoption with Data Standards• Current Pilot• Under Development• Future Plans

All Rights Reserved, Duke Medicine 2007

The Promise of Interoperability• Interoperability is based on common data standards and is a

pre-requisite for the aggregation and sharing of data• Interoperability in Life Sciences research can lead to faster

conclusions via analysis of shared/aggregated data • In health care, widespread interoperability can lead to better

outcomes in many areas ranging from high quality care, individual patient safety during treatment, to population safety from epidemics, to the everyday chronic and acute care of millions of citizens, irrespective of location

• Currently, vendor systems are not interoperable, institutional systems are not interoperable, and enterprises are not interoperable.

• Interoperability is the future

Published with permission

All Rights Reserved, Duke Medicine 2007

Background: The caBIG Initiative

• In 2003 Dr. Andrew von Eschenbach set the caBIG initiative in motion

• Sharing and expanding interoperable datasets for analysis - a crucial goal

• Accelerated research, cures and better outcomes• The success of this initiative relies on the incorporation of

standard datatypes, semantics, and software systems i.e. the semantic and syntactic - all working in harmony

• Interoperability is the future and caBIG leads the charge

All Rights Reserved, Duke Medicine 2007

Background: Interoperability Fundamentals• The fundamental unit of interoperability is the data standard (HL7

BRIDG, LS_DAM etc.)• The fundamental unit in caBIG data standards currently is the

Common Data Element (CDE) which uses ISO 11179 that incorporates standard semantics (vocabularies, terminologies etc.) and standard value domains/datatypes in it’s metadata (mapping of ISO 11179 to 21090)

• caBIG uses CDEs in Clinical/Life Sciences research/studies• In a Yin/Yang relationship (syntactic/semantic), software

encapsulates data and their CDEs (metadata) to facilitate data sharing - automated interoperability could be realized

• Other groups are creating their own data standards and are willing to use caBIG as a model

All Rights Reserved, Duke Medicine 2007

GROUPS OF STANDARDS

• GENERIC, BROAD USEXML, TCP/IP, Web services, OCL, CCOW, SECURITY, GIS, etc.W3C, IETF, OMG, HL7 others.

• DATA ELEMENTSRIM, DATA ELEMENTS, DATA TYPES, TERMINOLOGY, TEMPLATES, CDAHL7, CEN, ISO, NCPDP, X12N, IEEE, SNOMED. LOINC, RXNORM, SPL

• DATA INTERCHANGEStructured and free form documents, imagesHL7 V2.N AND V3, DICOM, IEEE X73, ASTM, NCPDP, X12n and others

• KNOWLEDGE REPRESENTATIONGuidelines and Protocols, ARDEN SYNTAX, GLIF, GEM, PRODIOGY, PROTIGÉ, vMR, GELLO, othersHL7, ASTM, UK, others

• ELECTRONIC HEALTH RECORDFUNCTIONAL REQUIREMENTS, CONTENT DATA SETS, EHR MODEL, CCR, TRANSFER EHR, other HERHL7, ASTM, CEN, openEHR

• APPLICATION LEVEL SUPPORTIDENTIFIERS, RESOURCE REGISTRIES, TOOL SETS, CONFORMANCE AND IMPLEMENTATION MANUALSHIPAA, HL7, ASTM, ISO, CEN

Published with permission

All Rights Reserved, Duke Medicine 2007

Standards at Duke• Duke is facilitating several projects in HL7

– Cardiovascular DAM– Anesthesiology DAM– InfoButton Standard– Tuberculosis DAM

• The caBIG Standards– ISO 21090/11179, – HL7 – Semantics - Vocabularies, Terminologies and

Ontologies (LOINC etc.)• Several applications with standards (listed later)

All Rights Reserved, Duke Medicine 2007

DATA

KNOWLEDGE

PROCESS

STA

ND

AR

DS

Low cost

Accessible

Patient Safety

Quality

Analysis

Statistics

Research

Clinical Trials

EBM

Analysis

Statistics

Research

Clinical Trials

EBM

Analysis

Statistics

Research

Clinical Trials

EBM

Analysis

Statistics

Research

Clinical Trials

Surveillance

Analysis

Public/PrivatePartnership

Vendor/ProviderPartnership

Published with permission

Privacy, Security, Trust, Integrity

All Rights Reserved, Duke Medicine 2007

PreventiveCare

Acute Care ChronicCare

Data Data Data

DecisionSupport

Decision Support Decision Support

Norm

al

Concern

Abnorm

al

Outpatient

Specialty

Em

ergency

Hospital

Intensive

Control

Treatm

ent

Perform

ance

Sym

ptoms

Diagnoses

Treatm

ent

Outcom

es

Genom

ic Medicine

Nursing Homes, etc. Home Care

Published with permission

A view of the healthcare world

All Rights Reserved, Duke Medicine 2007

Where do standards start?• With the smallest element – the data element• If we define a structured set of any and all data

elements that might be contained within an electronic health record, and if

• we include precise and unambiguous definitions, data types, units, roles and use, and many other attributes, and if

• we define unique value sets for these data elements (single, integrated terminology)

• then we can achieve interoperability independently of how these data elements might be packaged for interchange.

Published with permission

All Rights Reserved, Duke Medicine 2007

EHR Interoperability Diagram

PatientEncounter

Provider EHR

Database

PersonalEHR

PopulationEHR

Profile

Profile

ProfileResearchDatabase

EnterpriseData

WarehousePersonal

EHR

Profile

Disease Registry

Profile Billing/Claims

Institution EHR

Database

ResearchDatabaseResearch

Database

Disease RegistryDisease

Registry

Derived from master data element registry

Published with permission

All Rights Reserved, Duke Medicine 2007

caBIG Interoperability Development at Duke• Cancer Central Participant Registry (C3PR)• caTRIP

– Many elements are part of other caBIG applications

• RProteomics• Duke Cancer Proteome Characterization Center

(proposal)• VCDE/Architecture Mentorship

All Rights Reserved, Duke Medicine 2007

caBIG Adoption at Duke• Flagship C3D implementation

– Many trials in production, many more in development (all CDE based)

• caAERS (CALGB and Duke) (in progress)• C3PR multi-center pilot • caGRID• RProteomics (soon to be in development)• caArray

All Rights Reserved, Duke Medicine 2007

Integration Hub

CTMS(eResearch)

CDMSC3D

(Oracle Clinical)

Safety Desk(ExtraView)

S/AE Reporting(caAERS)

API

Web Service

Pull Study Status/Agentfrom eResearch

MedWatch input from e-mail, fax, paper,

caAERS

Local investigator assessment for IRB

routing determination

Local IRB and/or affiliate sites

(FDA for SAEs)

MedWatch form generation for

Duke held IND/IDE

Promised API

Clinical Connector

eIRB(Click

Commerce)

Protocol #Short Title

Investigator nameStudy Status (initial)

Data Entity Overview

All Rights Reserved, Duke Medicine 2007

Integration Hub

CTMS(eResearch)

CDMSC3D

(Oracle Clinical)

Safety Desk(ExtraView)

S/AE Reporting(caAERS)

API

Web Service

Pull Study Status/Agentfrom eResearch

MedWatch input from e-mail, fax, paper,

caAERS

Local investigator assessment for IRB

routing determination

Local IRB and/or affiliate sites

(FDA for SAEs)

MedWatch form generation for

Duke held IND/IDE

Promised API

Clinical Connector

eIRB(Click

Commerce)

Protocol #Short Title

Investigator nameStudy Status (initial)

CDE BASED

Data Entity Overview CDE Based

All Rights Reserved, Duke Medicine 2007

Program Project (P01) Proposal

Data Export Process

Invoice

Velos eResearchClinical Trial Management System ( CTMS )

Protocol/ Patient Management

BudgetReport/Financial

Data

Investigator/ CRC

Regulatory DocsBudget

Trial Protocol Dev .

DUKE Systems

caBIGTM

Applications

Invoice

Velos eResearchClinical Trial Management System ( CTMS )

Protocol/ Patient Management

BudgetReport/Financial

Data

Investigator/ CRC

Regulatory DocsBudget

Trial Protocol Dev .

DUKE Systems

caBIGTM

Applications

C3D Bladder (M-I)

Protocol

C3D BladderProtocol

Integration Hub

Reports

Analyses

Reports

Analyses

Via web service

iReviewTM

Query

Report Interface

iReviewTM

Query

Report Interface

Tool and Investigator

Report

Report

Report

Investigator

Report

Report

Report

caTissue

Web Service

Cancer Adverse Event Reporting System (caAERS)

ClinicalConnector

SAS Datasets

All Rights Reserved, Duke Medicine 2007

Program Project (P01) Proposal CDE Based

Data Export Process

Invoice

Velos eResearchClinical Trial Management System ( CTMS )

Protocol/ Patient Management

BudgetReport/Financial

Data

Investigator/ CRC

Regulatory DocsBudget

Trial Protocol Dev .

DUKE Systems

caBIGTM

Applications

Invoice

Velos eResearchClinical Trial Management System ( CTMS )

Protocol/ Patient Management

BudgetReport/Financial

Data

Investigator/ CRC

Regulatory DocsBudget

Trial Protocol Dev .

DUKE Systems

caBIGTM

Applications

C3D Bladder (M-I)

Protocol

C3D BladderProtocol

Integration Hub

Reports

Analyses

Reports

Analyses

Via web service

iReviewTM

Query

Report Interface

iReviewTM

Query

Report Interface

Tool and Investigator

Report

Report

Report

Investigator

Report

Report

Report

caTissue

Web Service

Cancer Adverse Event Reporting System (caAERS)

ClinicalConnector

SAS Datasets

CDE BASED

All Rights Reserved, Duke Medicine 2007

Duke Proteome Characterization Center (PCC)(Proposed)

Key: Red arrows indicate semantic involvement

All Rights Reserved, Duke Medicine 2007

Duke Proteome Characterization Characterization Center (PCC)(Proposed) CDE Based

CDE BASED

All Rights Reserved, Duke Medicine 2007

Benefits of caBIG Interoperability Adoption

• C3D– Top-shelf CDMS (with CBIIT tech. support)– Regulatory compliant (21 CFR Part 11)– Shared Global Library (parent-child)– caDSR-Direct (local curation ability)– Thousands of Common Data Elements– Development curve

• Less steep now, new trials up much faster• More cluster (brain, breast, etc) replication

All Rights Reserved, Duke Medicine 2007

Supported Projects on caBIG PlatformsGoal: All Investigator-initiated Clinical Trials (all CDE based)•Currently ~40 trials in production or development

– Single and multi-center– Both IND/IDE and Registry

•International– Launching first trial with Beijing Cancer

Hospital• …more to come

•Funding Mix – Gov’t – NIH (NCI, NINDS, etc) and DoD– Foundation (Komen, etc.)– Industry/Pharma

All Rights Reserved, Duke Medicine 2007

Supported Projects on caBIG Platforms CDE Based

Goal: All Investigator-initiated Clinical Trials (all CDE based)•Currently ~40 trials in production or development

– Single and multi-center– Both IND/IDE and Registry

•International– Launching first trial with Beijing Cancer

Hospital• …more to come

•Funding Mix – Gov’t – NIH (NCI, NINDS, etc) and DoD– Foundation (Komen, etc.)– Industry/Pharma

CDE BASED

All Rights Reserved, Duke Medicine 2007

Duke-Beijing Abraxane Trial CDE Based

First CT collaboration

caTissue-based tumor registry

Future network in China

CDE BASED

All Rights Reserved, Duke Medicine 2007

C3PR Multi-center Pilot (Completed) CDE Based

• PANVAC multi-center vaccine study– Grid-enabled– Secure– Accessible– CDE based

Dorian

GTS

CDS

webSSO

caXchange

Key

IdP

Study ServiceRegistration

Service

XYZ Cancer Center

CDE BASED

All Rights Reserved, Duke Medicine 2007

Under Development• caTissue (works with annotated samples)

– Pilot implementation underway– Potential for broad institutional adoption

• caArray – Installed (annotation tool is very useful)– caGrid enabled

• Laboratory Information System (GE – in use)– Uses LOINC

• Data warehouse– Implements a text analytic tool using Radlex

All Rights Reserved, Duke Medicine 2007

Under Development (con’t)• DISCERN (in use)

– HL7 is part of the eBrowser (DSR)– Potential for broad institutional adoption (available

to all Duke Researchers)• caArray

– Installed (annotation tool is very useful)– caGrid enabled

• Laboratory Information System (GE – in use)– Uses LOINC

• Data warehouse (in use)– Implements a text analytic tool using RadLex

All Rights Reserved, Duke Medicine 2007

Under Development (con’t)

• Cardiovascular Data Standards (DCRI)– DAM has ~ 1285 CDEs (HL7 conversion next)– Country wide adoption expected

• Bioconductor (off site caBIG compatible app.) – Genomic analysis (heavy use)

• GenePattern (off site caBIG compatible app.)– Genomic analysis (heavy use)

All Rights Reserved, Duke Medicine 2007

Future Plans

• C3D (all CDE based)– Streamline trial start-up with eCRF

Harmonization– Exchange data with caAERS– Complete integration with Velos eResearch

• Expanded use of caARRAY by DBSR– Annotations

• China (all CDE/Forms based)– Set-up local Suite environment in Beijing– Build-out multi-center Chinese network

All Rights Reserved, Duke Medicine 2007

Future Plans (con’t)

• Proteomic Center (proposed)– Streamline all data collection and analysis

with annotation engine, Tranche– Integration of RProteomics for analysis

• Duke will continue their involvement with HL7 standards

• Continued expansion of CRF creation (CDE based) by other groups – caDSR-Direct (local curation ability)

All Rights Reserved, Duke Medicine 2007

Summary

• Duke is very involved with data standards• Interoperability is the future• Currently, vendor systems are not interoperable, institutional

systems are not interoperable, and enterprises are not interoperable

• The fundamental unit of interoperability is the data standard (HL7 BRIDG, LS_DAM etc.)

• The fundamental unit in caBIG data standards currently is the Common Data Element (CDE) which uses ISO 11179 that incorporates standard semantics (vocabularies, terminologies etc.) and standard value domains/datatypes in it’s metadata

• For data standards, start with the smallest element – the data element

All Rights Reserved, Duke Medicine 2007

Acknowledgements

• DBSRWilliam Barry, DirectorPankaj Agarwal

• DTMIW. Ed Hammond, DirectorAnita Walden

• CCISBob Annecharico, DirectorMohammad Farid

• DCRIBrian McCourt, Assoc. DirectorRebecca Wilgus

All Rights Reserved, Duke Medicine 2007

Questions?

Q & A