Enabling Healthcare Terminology Interoperability using HDD … · 2015. 1. 29. · The 3M HDD is a...
Transcript of Enabling Healthcare Terminology Interoperability using HDD … · 2015. 1. 29. · The 3M HDD is a...
1
3M Health Information Systems
© 3M 2015. All rights reserved
Enabling Healthcare Terminology Interoperability using HDD Access
http://www.hddaccess.com
Biomedical Informatics Graduate Seminar
University of Utah
January 29, 2015
Senthil K. Nachimuthu, MD, PhD
3M Health Information Systems, Inc., Murray, Utah
2
3M Health Information Systems
© 3M 2015. All rights reserved
Conflict of Interest Disclosure
3M Health Information Systems, Inc.
― Medical Informaticist, Clinical Terminology
University of Utah School of Medicine
― Research Associate, Division of Epidemiology
Elected positions and Standards Development
― IHTSDO – US representative to Technical Committee
― Past Chair, AMIA Open Source Working Group
― Co-author, HL7 Common Terminology Services v2 (CTS2)
3
3M Health Information Systems
© 3M 2015. All rights reserved
Overview
Understanding terminology needs of healthcare organizations
The need for interoperability
Our approach to the interoperability question
Business model – publicly available and open source
HDD Access community
What’s next for HDD Access?
HDD Access demo
4
3M Health Information Systems
© 3M 2015. All rights reserved
In a typical hospital near you…
5
3M Health Information Systems
© 3M 2015. All rights reserved
Their computers often don’t speak the same languageHow do
you do?
Nimeni on
Linus.
நான் ஹெல்சிங்கியில்இருந்து
வருகிறேன்.
Είμαι 30
ετών.
Уровень глюкозы в
крови 90 мг / дл.
Radiografia del
torace non mostra
risultati anomali.
즉 2백달러될것입니다. 감사합니다!
6
3M Health Information Systems
© 3M 2015. All rights reserved
If they could all speak the same language…
How do you do?
Nimeni on Linus.நான் ஹெல்சிங்கியில் இருந்து
வருகிறேன்.
Είμαι 30 ετών.
Уровень глюкозы в крови
90 мг / дл.
Radiografia del torace non
mostra risultati anomali.
즉 2백달러될것입니다.
감사합니다!
How do you do?
My name is Linus..
I am from Helsinki.
I’m 30 years old.
My blood glucose is 90
mg/dL.
Chest X-ray shows no
abnormal findings.
That will be $200. Thank you.
7
3M Health Information Systems
© 3M 2015. All rights reserved
How Many Terminologies Does an Organization Need?
Standard terminologies
― Clinical: SNOMED CT
― Lab: LOINC
― Pharmacy: RxNorm, other
proprietary pharmacy
terminologies
― Billing/reporting: ICD-9-CM,
ICD-10-CM, ICD-10-PCS,
CPT, HCPCS, DRG, and
many others
Local terminologies
― Those created by vendors of
EMRs, pharmacy and lab
systems
― Those created by hospitals
themselves
― Often “uncontrolled” without
consistent governance
processes
8
3M Health Information Systems
© 3M 2015. All rights reserved
Our Answer to the Terminology Management Question
The 3M Healthcare Data Dictionary (3M HDD) was created in
the early 90s through an SBIR grant
Two of the early users were the DoD EHR and Intermountain
Healthcare
The 3M HDD is a concept-based terminology server that
integrates the current version of many standard and local
terminologies used by hospitals and other healthcare
organizations
Partitioning capabilities are included to support terminology
content from different organizations
9
3M Health Information Systems
© 3M 2015. All rights reserved
DoD, VA and HDD Access
DoD AHLTA EHR uses the HDD as its data dictionary
VA VistA EHR system uses its own terminology
In 2009, the DoD and the VA announced the joint iEHR
(Integrated Electronic Health Record) project
The DoD and VA gave 3M a contract in 2012 to make the 3M
HDD publicly available as HDD Access, and they will use
HDD Access for interoperability between their EHR systems
10
3M Health Information Systems
© 3M 2015. All rights reserved
What is HDD Access?
An open source terminology engine to enable
Implementation of various standard terminologies
Interoperability between multiple local and standard terminologies
Contains various standard terminologies and a subset of the 3M
HDD to enable interoperability between them
HDD Access supports standard terminologies by bridging the
gap between SDOs and users – it doesn’t compete with
standard terminologies, but it helps their implementation
11
3M Health Information Systems
© 3M 2015. All rights reserved
HDD Access Components
Includes terminologies
that we have permission
to release publicly
User can create local
extensions without
modifying the core
content
Database (Oracle, SQL
Server, MySQL, PostgreSQL)
Terminology browser,
authoring tool and search
engine
API (HL7 CTS, 3M)
Runs on Windows and Linux
Import/export tools
1500+ users
Discussion forum, blog, FAQ
Open Exchange – users can
share their contributions
Container
CommunityContent
12
3M Health Information Systems
© 3M 2015. All rights reserved
Components of the HDD
Controlled Medical Vocabulary
― Collection of clinical concepts
Knowledge Base
― Relationships among concepts
Medical Information Model
― Establishes clinical context
Decision Support
Alerts Reports
Knowledge Base
(Semantic Network)
Controlled Medical Vocabulary
(Medical Concepts)
Medical
Information
Model
*Slides 11-15 from Shaun Shakib, PhD
13
3M Health Information Systems
© 3M 2015. All rights reserved
A collection of medical concepts, organized to
support synonyms and other lexical
characteristics
• concept: a unique, definable idea or object that has a very
specific, known meaning
Sodium
Lab ResultLab Test Chem 4
Potassium
ChlorideGlucose
Concept Based Vocabulary
14
3M Health Information Systems
© 3M 2015. All rights reserved
COLD COLD COLDConcept
Representation
Concept
Domain
Concept
Definition
Concept ID
a sensory
perception
a pulmonary
diagnosis
an upper
respiratory
viral infection
“I’m feeling
cold”
Chronic
Obstructive
Lung Disease
“I have a
cold”
68215 1005480 1005313
Concept Based Vocabulary
15
3M Health Information Systems
© 3M 2015. All rights reserved
Knowledge Base/Relationships
Lab Test
Chem 4
Sodium
is-a
Lab Result
Potassium
Chloride
Glucose
is-component-of
16
3M Health Information Systems
© 3M 2015. All rights reserved
Medical Information Model
Allergy Type
Allergen
Severity
Reaction
Date
Allergy Event
Information Model
Patient Allergy
Drug (1550)
Aspirin (3000252119)
Mild (1554)
Rash (82559)
20000703
Instance Data
Type Allergen
Severity Reaction
Allergy Event
HDD
Date
20000703
Aspirin
Mild
Drug
Rash
Information Model - instantiated
17
3M Health Information Systems
© 3M 2015. All rights reserved
Point-to-Point vs Centralized Mapping
Epic
Siemens
HIE
CrimsonEDW
2
)1( nnmaps
LOINC
SNOMED
CT
RxNorm
Cerner
Epic
Cerner
RxNorm
Crimson
HIE
LOINC
SNOMED
CT
HDD
EDW
Siemens
(n-1) maps
HDD reduces mapping and maintenance efforts by using centralized mapping.
We add point to point mappings when necessary in a small number of cases.
18
3M Health Information Systems
© 3M 2015. All rights reserved
HDD Access Content v.29 – Standard Terminologies
ICD-9-CM Diagnoses
ICD-9-CM Procedures
ICD-10-CM
ICD-10-PCS (codes but not
attributes/hierarchies)
HCPCS Level II
HCPCS Modifiers
Terminologies that we have permission to include in HDD Access are supported and
duly acknowledged
MS-DRG
TRICARE/CHAMPUS MS-
DRG
APC
MS-DRG MDC
TRICARE/CHAMPUS MS-
DRG MDC
RxNorm
19
3M Health Information Systems
© 3M 2015. All rights reserved
HDD Access Terminology Design
Concept-based terminology, with concept permanence and
graceful evolution
Multi-hierarchical terminology
Supports semantic relationships
Supports and integrates terminology models and content of
multiple standard and local terminologies
Supports mappings between equivalent and non-equivalent
concepts
20
3M Health Information Systems
© 3M 2015. All rights reserved
HDD Access software is open source under the Apache
License v2 – source code is published in addition to the
binary installers
HDD Access content is “publicly available” under the HDD
Access Content License
― You cannot modify or re-release the core HDD Access content
― But you can extend it and release your extensions as you please
― This is done to protect the integrity of the terminology content, and
in turn, patient safety
HDD Access Licenses
21
3M Health Information Systems
© 3M 2015. All rights reserved
HDD Access Community
0
500
1000
1500
2000
Au…
Se…
Oc…
No…
De… Ja
…F
e…M
a…A
p…M
a… Ju…
Jul…
Au…
Se…
Oc…
No…
De… Ja
…F
e…M
a…A
p…M
a… Ju…
Jul…
Au…
Se…
Oc…
No…
NewUsers
Total Users
•More than 1,500 users, from government, industry (HIS vendors,
consulting), academia, and international users
•Active discussion forums, blog, FAQ
22
3M Health Information Systems
© 3M 2015. All rights reserved
HDD Access Content – Data Row Counts
0
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
3,500,000
4,000,000 Concepts
Relationshships
Representations
RepresentationContexts
23
3M Health Information Systems
© 3M 2015. All rights reserved
HDD Access Download Counts
0
50
100
150
200
250
Aug
-12
Sep
-12
Oct
-12
Nov
-12
Dec
-12
Jan-
13F
eb-1
3M
ar-1
3A
pr-1
3M
ay-1
3Ju
n-13
Jul-1
3A
ug-1
3S
ep-1
3O
ct-1
3N
ov-1
3D
ec-1
3Ja
n-14
Feb
-14
Mar
-14
Apr
-14
May
-14
Jun-
14Ju
l-14
Aug
-14
Sep
-14
Oct
-14
Nov
-14
Dec
-14
Total monthlyInstallerBinarySourceDB SourceContentDocumentation
24
3M Health Information Systems
© 3M 2015. All rights reserved
Enabling Interoperability
Local Extensions
Namespace Dependencies
Demo
25
3M Health Information Systems
© 3M 2015. All rights reserved
Local Extensions
Local extensions are implemented as “namespaces” where
the user can author their own terminology content without
modifying the core content
We provide a unique range of identifiers for each user’s local
extension to avoid collision with other users’ extensions
HDD Access Content License allows the user to use or share
their local extension with others
HDD Access tools allow export and import of local extensions
26
3M Health Information Systems
© 3M 2015. All rights reserved
Namespace Dependencies
HDD Access core content and local extensions are made of
“namespaces”, which are logical partitions of the content
Namespaces may have dependencies among them, since we
integrate multiple terminologies using a central concept-
based terminology
Example 1: The SNOMED CT concept of “Aspirin
(substance)” and the RxNorm concept of “Aspirin” (ingredient)
are mapped to the HDD concept “Aspirin” – the SNOMED CT
namespace and the RxNorm namespace depend on the core
namespace
27
3M Health Information Systems
© 3M 2015. All rights reserved
Namespace Dependencies
Example 2: If we load the “SNOMED CT to ICD-10-CM
Mappings from NLM” into HDD Access, the map set will be in
a separate namespace which depends on the SNOMED CT
namespace and the ICD-10-CM namespace
Soon, we will have a complex network of namespaces
The dependencies are between specific versions of
namespaces, rather than the namespaces themselves
The “versioned” namespace dependencies form a directed
acyclic graph
28
3M Health Information Systems
© 3M 2015. All rights reserved
Namespace Dependencies – Our Approach
Users define the namespaces that their namespace depends
on
The “versioned” namespace dependencies are automatically
calculated by the software
Namespace dependencies form a directed acyclic graph
Content can only be moved from the “leaf” towards the “root”
of the dependency tree, not in the other direction (which
would break concept permanence)
29
3M Health Information Systems
© 3M 2015. All rights reserved
Demo
Creating your own local extension
― Your namespace will be automatically created
Creating your own terminology
― Let’s create an example drug terminology
Creating your own mappings
― Let’s map your drug terminology concepts to RxNorm
30
3M Health Information Systems
© 3M 2015. All rights reserved
Thank you!
References
HDD Access website: http://www.hddaccess.com
Online browser: http://search.hddaccess.com
FAQ: http://www.hddaccess.com/home/faqs
Discussion forums: http://www.hddaccess.com/forum
Apache License v2: http://www.apache.org/licenses/LICENSE-2.0.html
HDD Access Content License: http://www.hddaccess.com/hdd-content-download-license
Contact us: http://www.hddaccess.com/contact
3M HDD: http://www.3mhis.com/terminology
Please send your questions to
Senthil K. Nachimuthu, MD, PhD
3M Health Information Systems, Inc.