Health care wants Linked Data, a semantic web of taxonomies and patient information that empowers patients, doctors and researchers. Hoot72, a straightforward way to break into the silo's of today, is a key step to get there.
2. Apple Pie
Patient Researcher
Linked Health Data
Doctor IT
Insurance Manager
E Pluribus Unum!
3. Where Are We?
• Health Data in “islands”, “bunkers”
• Admissions/Pharmacy/Lab ... Systems
• But the Systems do chat (internally)
• “Patient Arrival ...”, “Observation made ...”
• A stew of HL7 v2.x messages
• Billing data does leave the premises!
4. They speak v2
“HL7 version 2 is a major
breakthrough and market
success. More than 93%
hospitals in US are using this
standard” - Health Level
Horizon (HLH) Project
Source: Neotool, V3 vs V2
5. One “tweet” ...
MSH|^~&|REGADT|MCM|IFENG||199112311501||ADT^A04^ADT_A01|000001|P|2.4|||
EVN|A04|199901101500|199901101400|01||199901101410
PID|||191919^^GENHOS^MR~371-66-9256^^^USSSA^SS|253763|MASSIE^JAMES^A||
19560129|M|||171 ZOBERLEIN^^ISHPEMING^MI^49849^""^||(900)485-5344|
(900)485-5344||S^^HL70002|C^^HL70006|10199925^^^GENHOS^AN|371-66-9256||
NK1|1|MASSIE^ELLEN|SPOUSE^^HL70063|171
ZOBERLEIN^^ISHPEMING^MI^49849^""^
|(900)485-5344|(900)545-1234~(900)545-1200|EC1^FIRST EMERGENCY
CONTACT^HL70131
NK1|2|MASSIE^MARYLOU|MOTHER^^HL70063|300
ZOBERLEIN^^ISHPEMING^MI^49849^""^
|(900)485-5344|(900)545-1234~(900)545-1200|EC2^SECOND EMERGENCY
CONTACT^HL70131
NK1|3
NK1|4|||123 INDUSTRY WAY^^ISHPEMING^MI^49849^""^||(900)545-1200|
EM^EMPLOYER^HL70131|19940605||PROGRAMMER|||ACME SOFTWARE COMPANY
PV1||O|O/R||||0148^ADDISON,JAMES|0148^ADDISON,JAMES||AMB|||||||
0148^ADDISON,JAMES|S|1400|A|||||||||||||||||||GENHOS|||||199501101410|
PV2||||||||199901101400|||||||||||||||||||||||||199901101400
ROL||AD|CP^^HL70443|0148^ADDISON,JAMES
OBX||NM|3141-9^BODY WEIGHT^LN||62|kg|||||F
James was admitted ... his wife is his emergency contact ... here’s his weight ...
6. My Nurse, My Kin
“The HL7 we receive from the various sources systems can be wildly different
(some are not even near to being to spec). Additional we can receive the same
data (think heart rate) from multiple systems, each which represent this
differently.” -- System Implementer
Next of kin is a mother ...
NK1|1|MASSIE^ELLEN|MTH^Mother^HL70063||||EC1^FIRST EMERGENCY CONTACT^HL70131
Or a nurse!
NK1|1|Jones^Jane^Lee^^RN|VAB^Vaccine administered by (Name)^HL70063|
System-wide
HL7 v2 only “suggests” - too messy for^integration!
7. Disintegrated!
• Full integration costly
• custom models, brittle
• only stray outside when forced
• (Today’s) Web veneers
• for men, not machines
• Another Option? ...
8. HL7 v3 - the Fix?
• HL7 v3 - Object-based
• End 2005, a do over, ten years in the making
• Object Model (RIM), not V2-compatible
• Bonus: XML goodness
• Beyond Interface - Common Model
• Modern systems to be “RIM-Inside”
• Remake all systems
9. But ...
“Frozen interface syndrome: existing HL7
message interfaces are pretty much stuck at 2.2.
Meaning that application vendors don't like to
change for change's sake -- there needs to be a
business motivation.”
-- Wes Rishel, Gartner Group
• Government Mandate (UK), some green field
• Just another opportunity to gateway
• Just polite talk? Are we stuck?
10. Back Inside the Stew
MSH|^~&||GA0000||VAERS PROCESSOR|20010331605||ORU^R01|20010422GA03|T|2.3.1|||AL|
PID|||1234^^^^SR~1234-12^^^^LR~00725^^^^MR||Doe^John^Fitzgerald^JR^^^L||20001007|M||
2106-3^White^HL70005|123 Peachtree St^APT 3B^Atlanta^GA^30210^^M^^GA067||(678)
555-1212^^PRN|ORC|CN|||||||||||1234567^Welby^Marcus^J^Jr^Dr.^MD^L|||||||||Peachtree Clinic|101
Main Street^^Atlanta^GA^38765^^O^^GA121|(404) 554-9097^^WPN|101 Main
Street^^Atlanta^GA^38765^^O^^GA121|OBR|1|||^CDC VAERS-1 (FDA) Report|||20010316|OBX|1|
CE|30955-9&30956-7^Vaccine type^LN|1|08^HepB-Adolescent/pediatric^CVX|OBX|2|CE|
30955-9&30957-5^Manufacturer^LN|1|MSD^Merck^MVX|OBX|3|ST|30955-9&30959-1^Lot
number^LN|1|MRK12345|OBX|4|CE|30955-9&30958-3^Route^LN|1|IM^Intramuscular^HL70162|
OBX|5|CE|30955-9&31034-2^Site^LN|1|LA^Left arm^ HL70163|OBX|6|NM|
30955-9&30960-9^Number of previous doses^LN|1|01I
• Stuff and Codes, Codes, CODES!
• Messages (v2 or v3) just “trucks”
• If Codes a mess then ...
11. State of the Codes
• For substances, procedures, diseases ...
• Local, National, International; Care vs Billing
vs ...
• SNOMED CT, MeSH, ICPC, ICD-9-CM, DRG,
MDC, LOINC, CPT-4, HCPCS Level II, ICD-10,
HL7 Vocabulary Domains ...
• Too many, unwieldy, overlapping ...
• Extend, (re)structure, link
12. “In” Schemes
Scope Size Grain C/E*
SNOMED All 350K Varies C
LOINC Labs 50K? High C
NDC Drugs 11 Digits High E
ICD9-CM Diseases 10K? Low*** E
UMLS All** Combined Varies Varies
* Composed or Enumerated
** Assumes diversity. UMLS is 143 others: “not an attempt to build a single standard biomedical vocabulary.” RxNORM is its drugs.
*** “Other” is a catch-all, mandated for medicare billing
13. Focus: Coverage
• Problem List to SNOMED: “SNOMED CT ... can
exactly represent a large portion (92.3%) of the
terms” -- Mayo Clinic
• Local Lab Codes to LOINC: “Almost 19% of
LOINC codes are ‘other’” -- CDC
• LOINC to SNOMED: “Automatic mapping of
laboratory procedures from LOINC to SNOMED
CT remains incomplete and unsatisfactory” --
Mapping Issues, Olivier Bodenreider, MD, PhD
14. Elephant: Equivalence
there may be several different ways to express the same concept. Human users may be
able to recognise that these are essentially the same, but the rules for doing so must be
made explicit to be usable by computer.
-- Why is Terminology hard?, Alan Rector
• Finding vs Observable
• [serum_potassium_elevated_code] vs [Code: serum_potassium_code Value: elevated]
• Many and one
• “head injury” and “no intracranial bleed” vs “head injury without intracranial bleed”
• Little/No Lexical match
• “colon cancer” vs “malignant neoplasm” plus “primary site: colon”
• Beyond Equivalence - kind-of
• BrCA Breast Cancer, Cystic Fibrosis ... are Diseases-linked-to-Genes
15. Where’s the Ontology?
• Ontology: “an implementable model of the
entities that need to be understood in common in
order for some group of software systems and
their users to function and communicate at the
level required for a set of tasks” -- Alan Rector
• Code schemes slight/none: “deficiencies in the
knowledge representation languages used”
• Enter w3c’s OWL == Web Ontology Language
• Focus on meaning referenced by codes
17. Joining the “Semantic Web”
• Web of documents to Web of data
• Reuse: HTTP, URIs
• Add: Query (SPARQL), Represent
(RDF), Meaning (OWL)
• Silo’ed to Meaningful Linked Data
18. W3c HCLS - Power of linkage
• Gap: Trial Criteria, Patients
• Patients taking “Weight Loss Drugs”
• Patient NDC codes: DESOXYN TABLETS
(00074337701) ...
• Linked by Ontologies
• "may_treat" “Obese” to RxNORM
• RxNORM links in NDC codes
• Trial meets Patients in two steps
19. What about the Patient?
OR|20010331605||ORU^R01|20010422GA03|T|2.3.1|||AL|
725^^^^MR||Doe^John^Fitzgerald^JR^^^L||20001007|M||
2106-3^White^HL70005|123 Peachtree St^APT
3B^Atlanta^GA^30210^^M^^GA067||(678)
555-1212^^PRN||||||||||Peachtree Clinic|101 Main
Street^^Atlanta^GA^38765^^O^^GA121|(404)
554-9097^^WPN|101 Main
Street^^Atlanta^GA^38765^^O^^GA121|
Concepts linking Shouldn’t patients join in?
20. E Pluribus Unum ...
Patient Researcher
Linked Health Data
Codes
“Link-
Doctor
Maker”
Patients
Insurance Manager
21. Avoid Temptation!
• V2’s Yuck Factor - “Legacy”, “Messy”
• Formalism likes its fellows
• wait until V2 disappears?
• mate its successors to this Web
• Hold your noses. Go where the data is!
22. Enter Hoot72
• Hoot (from OWL), 72 (HL7 v2)
• Mapping: message to graph assertions
• pre-defined ontology small
• Automatic: drive off message definitions
• Presumes little so generally applicable
• Creative Commons License
23. Observation (OBX)
PID|||1234^^^^SR~1234-12^^^^LR~00725^^^^MR||Doe^John^Fitzgerald^JR^^^L|
...
OBX|4|CE|30949-2^Vaccination adverse event outcome^LN|1|H^required
hospitalization^NIP005|
type
Patient
context
personName
observation observationValue
familyName
givenName Doe
CodingSystem Code
middleName
Code CodingSystem
John Fitzgerald
LN 30949-2 H NIP005 Identifiers and Time not shown
24. A first step: Expose
• A Site graph == Here’s your
Hospital/Clinic
• Linking to do (to code ontologies ...)
• “Predicates” to interpret
• Reports/views to generate
• Inaccuracies to find
25. From Many, One ...
• Health Care unlinked today
• Linked Health Care = site graphs +
interacting ontologies, linked
• Hoot72, a key step. www.hoot72.org.