New Desiderata for Biomedical Terminologies

1
New Desiderata for Biomedical Terminologies

• Discussant
• James J. Cimino
• Columbia University

2
Dilemmas of Biomedical Terminologies
3
Dilemmata of Biomedical Terminologies

• Abstract concepts vs. real world entities
• Knowledge about concepts vs. knowledge about
  extensions in reality
• Terms with context-specific meanings
• Changes in knowledge vs. changes in meaning

4
Abstract Concepts vs. Real World Entities
Concept
5
Desiderata for Controlled Medical Data

• I - Capture what is known about the patient
• II - No information loss
• III - No false implications
• IV - Support retrieval
• V - Support reuse
• VI - Support aggregation
• VII - Support inference

6
Desiderata forControlled Medical Terminologies

• Provide identifiers for meanings we want to apply
  to the patient
• Make sure the semantics are universally
  understood, separate from linguistics
• Make sure that, as our understanding changes,
  original meaning is not forgotten
• Provide a bridge between what we record and how
  we reason

7
Patient Data as Extensions of Concepts

• Everything we record is an abstraction
• Context tells us how the abstractions are
  interpreted
• Example contexts of diagnoses in medical records
• Example events
• Example evolution of patient problems

8
Everything We Record is an Abstraction

• We dont say that concepts cause concepts or
  terms cause terms
• Nor do we say that we are storing bits of
  pneumonia in the EMR
• Pneumococcal pneumonia is something we know
  about that we associate with the patient through
  interpretation of what we know about the
  patient
• So is the infiltrate seen on the chest x-ray
• So is the chest x-ray

9
Context in the EMR What We See

• Lab results Sodium 130
• Problem list Hyponatremia
• Medication list Hypertonic Saline
• Medication administration Hypertonic Saline

10
Context in the EMR What We Record

• At lttimestampgt lab reports ltproceduregt with id
  ltIDgt and value ltvaluegt for ltpatientgt
• At 10AM EST May 1, 2005 lab reports Stat Serum
  Sodium with id 123123 and value 130 for Mr.
  Jones
• At lttimestampgt ltcliniciangt interprets ltIDgt as
  indicating ltconditiongt for ltpatientgt
• At 1001AM EST May 1, 2005, Dr. Brown interprets
  123123 as indicating Hyponatremia for Mr. Jones

11
Context in the EMR What We Record

• At lttimestampgt ltcliniciangt orders pharmacy item
  ltformulary itemgt with order id ltIDgt for ltpatientgt
• At 1002AM EST, May 1, 2005 Dr. Brown orders
  pharmacy time ltHypertonic Salinegt with order id
  3233 for Mr. Jones
• At lttimestampgt pharmacy delivers ltinventory itemgt
  with inventory id ltIDgt for order id ltIDgt for
  ltpatientgt
• At 1003AM EST, May 1, 2005 pharmacy delivers
  Lilly Product 5505 with inventory id 223232 for
  Mr. Jones

12
Context in the EMR What We Record

• At lttimestampgt decision support system suggests
  ltconditiongt for ltpatientgt
• At 1004AM EST, May 1, 2005 decision support
  system suggests Factitious Hyponatremia for Mr.
  Jones

13
Concepts or Entities?

• Concepts included in EMR
• Stat Serum Sodium
• Hyponatremia
• Hypertonic Saline
• Lilly Product 5505
• Factitious Hyponatremia
• Entities in Real World
• Mr. Jones
• Mr. Joness Blood
• Serum Specimen 55555
• Lab
• Test ID 123123
• Analyzer 114
• 130 (reading on analyzer)
• Dr. Brown
• Pharmacy
• Lilly Product 5505 223232
• Decision Support System

14
What Do Knowledge Do We Need?

• Stat Serum Sodium measures Sodium
• Hyponatremia is defined as Sodium below normal
  range
• Normal Range of Sodium is 135-145
• Lilly Product 5505 is a Hypertonic Saline
• 223232 is an available instance of Lilly Product
  5505
• Analyzer 114 uses Flame Photometry Method
• Factitious Hyponatremia occurs when Sodium is
  low, method is flame photometry, and
  triglycerides are high

15
How Do We Cope with Changes in Our Knowledge
About the Patient?

• Barrrows RC, Johnson SB A data model that
  captures clinical reasoning about patient
  problems. SCAMC 1995402-405.
• Chest pain ? R/O CAD ? Costochondritis ?
  Resolved (S/P Costochondritis)
• Thread of assessments at points in time
• What do you think the patient has vs. (please
  tell the court) What did you think the patient
  had?

16
How Do We Cope with Changes in Our General
Knowledge?

• Change in
• what tests measure
• formulary
• normal ranges
• definition of disease
• The test if we change update the knowledge, do
  we change the old data?

17
Referent Tracking

• We do this now (or we should)
• Implicit vs. explicit interpretation
• Each referent still has to correspond to some
  meaning
• Solution must be practical, usable, and
  maintainable

18
The Desiderata Revisited

• Concept orientation - what is the alternative?
• Concept permanence and graceful evolution -
  version control
• Formal definitions - add to knowledge vs.
  recognize change
• Reject NEC - store what the patient has and
  classify later
• Multiple granularities - patient level vs. reuse
• Representing context - the implicit meaning in
  the EMR design

19
General Philosophy of Concepts and Data

• Concept plus Context
• Meaning plus Stamp

Recapitulates
Philately
Ontology