Clinical Observations Interoperability: EMR + Clinical Trials
Abstract
There is great perceived value in being able to work with clinical research data and clinical care data across clinical domains, e.g. health care, pharmaceutical development, pharmcosurveillance. This include the interest to re-use patient data collected in the care context (from EMRs) for data collection points in a clinical research protocol.
Although there is significant overlap in the clinical data content of collected in clinical research and clinical care, the different emerging standard data representations are not taken that into consideration, therefore usage of data across domains is a challenge.
Clinical research data and clinical care data from EMRs have different information models and different terminological standards. CDISC's Study Data Tabulation Model (SDTM), an emerging standard for regulatory reporting of clinical research data, defines the study variables for a given clinical trial in terms of domains and observations. Although many EMRs do not use standards in the modeling of clinical information, there is a large body of work in the Healthcare/Clinical Practice arena where detailed clinical models (DCM) have been developed, e.g. HL7 CDA templates.
The goal of this task is to investigate, and if possible propose, a model that enable re-use of common observation models across the clinical trials and clinical practice contexts. This of course, include how Semantic Web specifications such as RDF/OWL can be leveraged to achieve this.
Task List/Project Plan
- Use Case Scenario with sample Clinical Research Protocols: Rachel
- 10 Clinical Research Protocols dealing with Labs, Patient Measurements and Diseases/Diagnoses: Rachel
- Functional Requirements Document: Vipul
Functional Requirements Version 1 with example content and mappings
Please e-mail Vipul Kashyap if you need write access to this.
- Clinical Data and Information Models based on Data/Functional Requirements
- Clinical Trial Data and Information Models based on Data/Functional Requirements
- Clinical Practice Ontology
- Clinical Trials Ontology
- Computer-based Patient Record Ontoloty
- SPARQL queries
- Using N3 Rules for Clinical Trial Eligibility
- Mappings using N3 Rules
- Mappings across Clinical Trials and Clinical Practice Ontology
- Mappings across vocabularies relevant to the protocol - e.g., Snomed, MedDRA: TBD
- Propose "computable specifications" using OWL 1.1, Rules or both: TBD
- Identify and propose strategies to get access to real world data, esp. healthcare: Jyoti Pathak
Face to Face November 2007
Tentative Agenda for F2F in November 2007
Minutes
Participants
- Stan Huff, Medical Informatician, Intermountain Healthcare
- Aaron Kamauu, RemedyMD
- Jerry Potrado, Roche
- Charlie Barr, Roche
- Amit Sheth, Wright State University
- Joey Coyle, Intermountain Healthcare
- Jyotishman Pathak, Mayo Clinic
- Rachel Richesson, South Florida University
Bo Andersson, AstraZeneca
Kerstin Forsberg, AstraZeneca
- Chimezie Ogbuji, Cleveland Clinic
- Yan Heras, Intermountain Healthcare
- Steve Dobson, Pfizer
- Vijay Bulusu, Pfizer
- Susie Stephens, Eli Lilly
- Miguel Valenzuela, Roche
- Tom Oniki, GE Healthcare
- Vipul Kashyap, Partners Healthcare
- Jennifer Fostel, NIEHS, NIH
- Landen Bain, CDISC
- Dan Corwin, Lexikos and SemantxLS
- Eric Neumann, Clinical Semantics Group
- Helen Chen, Agfa Healthcare
- Dan Russler, Oracle
- Matt Moores, Oracle
- Holger Stenzhorn, DERI
Use Case Scenarios
- Rachel Richesson: Use Case Scenario for Screening Patient Populations for Clinical Trial Eligibility
- Incorporate Stan Huff's suggestions for use case ... enumerated below:
- Doing mammography
- see nodules or calcifications
lead to testing ==> breast biopsies
- estrogen receptor results
- lump was benign or malignant
- underlying genetics on BRCA1 gene
- correlate the genetic makeup with phenotypic observations
- ... and so on ...
Narrow down Choices for Problems/Diagnoses (see UsecaseBackground-EMR-CT-Interop-8-27-07.pdf)
- Susie Stephen: Use Case Scneario based on known drugs in the market
- Need to coordinate with Rachel at some point.
- Jerry Potrado: Propose a narrowly scoped Clinical Trial Design that involves the above examples
- Miguel Valenzuela: Propose a use case scenario based on 21 CFR 11 Regulgations that involves the above examples
Landen Bain: CDISC High Level Use Case
Dan Corwin: Core Requirements for a simple matching web service
Dan Corwin: EMR Wrappers a way to install matching predicates on an EMR system
Dan Corwin: Search Agents how to automate searches for protocol candidates
Dan Corwin: : COI Architecture proposed for a prototype use case web app
Ronan Fox: Use Case Scenario for Clinical Trial Patient Recruitment
Teleconferences
Resources
Parallel Universes? Technology in Healthcare and Clinical Development - Paul Bleicher
Integrating EHR with EDC: When Two Worlds Collide - Paul Bleicher
Medical Informatics Perspectives on Leveraging the Electronic Medical Record in Pharma
Implementation of counted inclusion + exclusion criteria in OWL
IBM's SHER Usecase: Matching Patient Records to Clinical Trials Using Ontologies
The Future Vision of Electronic Health Records as an eSource for Clinical Research
In collaboration with researchers at Columbia University Medical Center) SHER has been used to find electronic patient records that match clinical trials criteria. Pellet, and Fact++ is used in conjunction with Scalable Highly Expressive Reasoner (SHER) as the mechanism for identifying cohorts (amongst 217K+ patients ). SNOMED-CT is used (an expressive Description Logic: EL++) as the ontology for query answering and justification.