Intelligent Terminologies to Support System Interfaces: The Medical Entities Dictionary Presentation to Guidant Corporation April 15, 2002

1
Intelligent Terminologies to Support System
InterfacesThe Medical Entities
DictionaryPresentation to Guidant
CorporationApril 15, 2002
2
Overview

• What is the MED?
• Support for Clinical Information Systems
• Support for Data Reuse
• MED Tools
• Knowledge-based Maintenance

3
New York Presbyterian HospitalClinical
Information Systems Architecture
4
Medical Entities Dictionary A Central
Terminology Repository
5
Communicating Terminology Changes
6
Solution Hierarchical Integration
K1
K2
7
MED Structure
Medical Entity
CHEM-7
Plasma Glucose
8
The MED Today

• Concept-based (70,000)
• Multiple hierarchy (90,000)
• Synonyms (170,000)
• Translations (120,000)
• Semantic links (130,000)
• Attributes (160,000)

9
Support for Data Reuse

• Generated by one system, used by another
• Different granularities (lumpers and splitters)
• Different semantics (impedance mismatch)

10
Translations with the MED
Injectable Gentamicin
Serum Gentamicin Level
Gentamicin
Gentamicn Sensitivity Test
Gentamicin Toxicity
11
Summary Reporting
Chem20 Display
Serum Glucose Test
Fingerstick Glucose Test
Plasma Glucose Test
12
DOP Summary
13
WebCIS Summary
14
Supporting Clinical Research

• Epidemiology - symptoms, incidence, natural
  history of disease
• Outcomes - effectiveness of therapy, ideal length
  of stay
• Recruitment - identifying eligible participants

15
Patient Recruitment

• Study of bisphosphonates in hypercalcemia
• Potential subjects treated before enrollment
• Alert checked for elevated calcium and sent
  message to research fellow
• Enrollment was complete in two months

16
Linking to Expert Systems
17
Linking to Expert Systems
18
Linking to Expert Systems
19
Terminology and Automated Decision Support

• Data monitor checks for triggering conditions
• Medical Logic Modules decide if warning
  conditions are present
• Message sent to appropriate channel
• Example Tuberculosis culture result

20
Decision Support Example TB

• Monitors for delayed culture results
• Sends message if result not equal to the code No
  growth
• One day, dozens of alerts about positive results
  but no organism was reported
• What happened?

21
How the Lab Fooled the Alert

• Alert looked for results No Growth
• Lab started reporting No Growth to Date
• No Growth to Date ? No Growth
• Solution Use the controlled terminology to map
  all No-Growth-like lab terms into a single class,
  and have the alert logic refer to the class.

22
How We Outsmarted the Lab(Before)
Medical Logic Module
No Growth to Date
No Growth
23
How We Outsmarted the Lab(After)
Medical Logic Module
No Growth to Date
No Growth
24
Smarter Retrievals from the Record

• Repository stores events and results
• Clinical problems at a different level of
  granularity
• Re-use knowledge to map from problems to clinical
  data
• Produce problem-specific views of the medical
  record

25
Concept-oriented (Heart)
Radiology 2/28/96 Head CT
Lab 12/28/96 Sickle Cell Test
Admission 3/14/96 Stroke
Lab 1/1/99 Blood Type Test
Radiology 2/1/97 Knee X Ray
Admission 2/14/98 Angina
Discharge 1/15/99 CHF
Radiology 2/23/99 Chest X Ray
Lab 1/1/99 Cardiac Enzyme Test
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
Linking to On-line Resources with Terminology

• Reviewing reports will generate information needs
• On-line information sources can satisfy that need
• Data from report can be used to automate the query

30
(No Transcript)
31
(No Transcript)
32
(No Transcript)
33
Linking Text Reports to On-line Information
Sources

• Natural Language Processing
• Data representation to support reuse
• Codification of information needs

34
(No Transcript)
35
(No Transcript)
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
(No Transcript)
41
MED Tools

• MUMPS MED Editor
• qrymed
• accessmed
• Web Browsers

42
(No Transcript)
43
MED Browsers
44
(No Transcript)
45
Vocabulary Construction Issues

• Understanding
• Modeling
• Creation
• Maintenance

46
Knowledge-based Maintenance

• Theory
• "A knowledge-based approach to vocabulary
  representation will improve maintenance and
  utility."

47
Adding New Terms

• Identify redundant terms
• Put new terms into existing classes
• Create new classes where appropriate

48
Put Terms into Existing Classes

• Theory The attributes of new terms can be used
  to identify classes
• Practice "Pushing" Terms

49
Pushing a Term
Medical Entity
Chemical
Laboratory Test
Carbo- hydrate
Bioactive Substance
Stat Glucose Test
Chemistry Test
Plasma Glucose Test
Glucose
Chem-7 Glucose Test
Chem-20 Glucose Test
50
Pushing a Term
Medical Entity
Chemical
Laboratory Test
Carbo- hydrate
Bioactive Substance
Stat Glucose Test
Chemistry Test
Plasma Glucose Test
Glucose
Stat Glucose Test
Chem-7 Glucose Test
Chem-20 Glucose Test
51
Pushing a Term
Medical Entity
Chemical
Laboratory Test
Carbo- hydrate
Bioactive Substance
Stat Glucose Test
Chemistry Test
Plasma Glucose Test
Glucose
Stat Glucose Test
Chem-7 Glucose Test
Chem-20 Glucose Test
Stat Glucose Test
52
Create New Classes

• Theory Attribute patterns can be detected which
  identify potential classes
• Practice Recursive partitioning of existing
  classes

53
Finding a New Class
Medical Entity
Laboratory Test
Chemical
Chemistry Test
Antigen
Core Antigen
HBC
Hepatitis B Core Antigen
54
Finding a New Class
Medical Entity
Medical Entity
Laboratory Test
Laboratory Test
Chemical
Chemical
Chemistry Test
Chemistry Test
Antigen
Antigen
Core Antigen
HBC
Hepatitis B Core Antigen Test
Hepatitis B Core Antigen
Hepatitis B Core Antigen
Core Antigen
HBC
55
Maintenance Tasks

• New Vocabularies (Laboratory)
• Changing Vocabularies (Pharmacy)

56
New Vocabulary Laboratory

• Original lab 2533 terms
• New lab 5291 terms
• Vocabulary delivered June 15, 1994
• Go live date July 24, 1994

57
Changing Vocabulary Pharmacy

• Started with 2091 drugs
• In two years, added 1827 drugs
• Classification by
• Ingredients
• AHFS Class
• Allergy
• DEA
• Form

58
Automated Classification
Medical Entity
Allergy Class
Chemical
Drug
Sulfa Allergy "S1"
Trimethoprim Allergy "65"
Antibiotic
Pharmacologic Substance
Trimethoprim/ Sulfamethoxizole Preparations
Sulfameth- oxizole
Trimeth- oprim
Bactrim "S1", "65"
Septra "S1"
59
Formulary Correction Statistics

• Among original 2091 drugs
• 334 unclassified drugs assigned classes
• 289 drugs assigned multiple classes
• 173 drugs discovered to be missing allergy codes
• Among additional 1827 drugs added
• 25 unclassified drugs assigned classes
• 121 drugs assigned multiple classes
• 38 drugs discovered to be missing allergy codes

60
Semi-Automated Maintenance

• Read formulary file
• Identify new drugs
• Link new drug to ingredient(s)
• Suggest classifying in preparation class
• Add new drug as per human reviewer

61
Interactive Classification
Adding "LASIX 20MG TAB" Generic Ingredient
"FUROSEMIDE" AHFS Class "DIURETICS" Add to
"FUROSEMIDE PREPARATION"? y Adding "ZAROXOLYN
5MG CAP" Generic Ingredient "METOLAZONE" AHFS
Class "DIURETICS" Add to "DIURETICS"?
n Create METOLAZONE PREPARATION" Class? y
62
Impact of "Theory into Practice"Better
management

• Easier to merge new vocabularies
• Easier to automate change management
• Higher quality through better modeling

63
Impact of Better ManagementMore Useful
Vocabulary

• MED is up-to-date for ancillary systems
• Easier to find terms in the MED
• Support for multiple conceptual levels
• More accurate database queries

64
Knowledge-Based Terminology Maintenance

• Forcing explicit definitions elicits meaning
• Explicit definitions force consistency
• Inferencing possible to
• automate classification
• identify new classes
• But
• Modeling is hard
• Agreement on definitions is hard

65
The Columbia Experience

• Capturing data from ancillary systems
• Natural language processing
• Reusable data summaries
• Merging data across systems
• Support for clinical research
• Linking to expert systems
• Automated decision support
• Smarter retrievals from the record
• Linking to on-line resources
• Terminology management