Data Normalization
Milestones
Data Normalization
 Goals
– To conduct the science for realizing semantic
interoperability and integration of diverse data
sources
– To develop tools and resources enabling the
generation of normalized EMR data for
secondary uses
Data Normalization

Target
Value Sets
Information
Models
Normalization
Targets
Tooling
Raw EMR
Data
Normalized
EMR Data
Normalization
Process
SHARPn Data
Normalization Team
PIs: Dr. Chris Chute; Dr Stan Huff
Program Management Team
– Manager: Lacey Hart; Assistant: Erin Martin
Members
Hongfang Liu (Lead)
Calvin Beebe (co-Lead)
Tom Oniki (co-Lead; Modeling Lead)
Harold Solbrig (CTS2 Lead)
Vinod Kaggal (Implementation Lead)
Kyle Marchant (CEM DB Lead)
Joey Coyle (Modeling)
Teresa Conway (Modeling)
Nathan David (Modeling)
Edinardo Potich (Modeling)
Cyndalynn Tilley (Modeling)
Les Westberg (CEM DB)
Kevin Peterson (CTS2)
Siddhartha Jonnalagadda
(Pipeline)
Dingcheng Li (Pipeline)
Modeling -Accomplishments
 CORE and Secondary Use
Models/terminology patient, meds, labs,
administrative dx and procedure,
disease/disorder
 Definition of value sets in terms of
standards where possible
Modeling - Accomplishments
Request site (in pilot)
CEM Browser
Modeling - Future work
Request site full launch
Browser launch
Browser enhancements (terminology
integration, knowledge repository
integration)
Better documentation
Model enhancements
Tooling – explore collaborations
Value Sets- Accomplishments
 SHARP Value Sets published as CTS2
Resolved Value Sets
– ICD-9
– RxNorm
– LOINC
– ECIS
– SNOMED-CT
Value Sets- Accomplishments
 CTS2 Services Created For
– SNOMED-CT Simple Refsets and Dynamic
Value Set Definition
– ECIS Value Sets and Maps
 CTS2 Participation in CIMI effort
Value Sets - Future work
 CTS2 Service for RxNorm and UMLS
access
– Allow creation of new value sets
– Links to concept descriptions and
relationships
 Integrate BioPortal Value Set Services
– Bridge Bioinformatics Ontology / Clinical
Terminology
Value Sets – Future Work
 Complete ECIS Server
– Will aid in CEM to ADL migration
 CTS2 Services available to CIMI project
 Integration
– Create Pipeline specific services coupled directly to
CTS2
– Tighter coupling of Concept Domains and Value Sets
– Ajax Widgets for viewing, integrating and Authoring
Pipeline - Accomplishments
Architecture revision to make pipeline
highly configurable
– Enable seamlessly pipeline integration of
components from different data sources (e.g.,
NLP)
Generic pipeline components
– XML Initializer
Pipeline - Accomplishments
Highly configurable syntactic and semantic
mappings
– Allow four kinds of syntactic mapping –
constant, 1:1, 1:n, and inference
Two options to process data
– XML tree-based (achieve semantic mapping
through UIMA annotations)
– Template based (achieve translation directly
in CEM annotations)
Pipeline – Future work
 Integrate NLP normalization pipeline into
data normalization
 GUIs to simplify the configurations of the
pipeline
– syntactic mapping
– Semantic mapping
Integrate the two options provided to
process data into a single project
CEM to DB - Accomplishments
Completed the following CEM to DB Mirth
Channel Development
– Patient CEM Channel (Add and Update)
– NotedDrug CEM Channel (Add)
– AdministrativeDiagnosis CEM Channel (Add)
– Lab (Quantitative, Narrative, Ordinal, and
Titer) CEM Channel (Add)
CEM DB - Accomplishments
Completed CEM DB Design and
Development
– Patient CEM Tables
 Demographics including Names, Addresses,
Telecom, and Language
 Patient Cross Reference and External IDs
– Clinical CEM Tables
 Index Data
 Source Data
 Patient Data
HL7 to XDR Channel Accomplishments
Completed the following HL7 to XDR
channels for data exchange
– HL7 AdminDX to XDR Channel (Sender)
– HL7 Lab to XDR Channel (Sender)
– HL7 Noted Drug to XDR Channel (Sender)
– HL7 Message Separator (Sender)
– HL7 Receive Message Separator (Receiver)
– ReceiveXDR and Extract HL7 (Receiver)
Next Steps for Channels and CEM
DB
Next Steps
– Complete the Update feature for Clinical CEM
Channels
– Enhance the Error Handling and resending of
error Message for channels
– Additional Relational field data storage for
Clinical CEM Messages
– Support CEM changes and new data types
In summary
 Continue effort in tooling for
– Modeling (through CEMBrowser and Request
Site)
– Value sets (through CTS2)
– Pipeline set up (GUIs for configuration and
tools for generating semantics mapping)
– CEM XML -> CEM DB, CEM RDF, etc