COAS Submission

1
COAS Submission

• 3M
• Care Data Systems
• CareFlowNet
• HBO C
• Philips Medical Systems
• Protocol Systems

2
Supporters Contributors
3
Outline

• Overview
• Scope of COAS
• Background on the submitters
• Usage examples
• Issues list - status and discussion

4
History

• June, 1997 - RFI issued
• Aug Sept, 1997 - Responses were received
• Dec., 1997 - RFP issued
• March, 1998 - Submitters started discussions
  (after HIMSS)
• April, 1998 - LOI deadline
• May, 1998 - Submission deadline
• Submission caveats
• Early effort was to identify issues and usage
  scenarios
• Straw-man IDL was based on a Protocol
  implementation to access vital signs.

5
Environment
Back-ends
Middleware
Lab results
User
Front-end
Reports
COAS interfaces? Middleware Server
Back-end Server Both
Images
6
Definitions

• Clinical Observations
• Any measurement, recording, or description
  WHAT of the anatomical, physiological,
  pathological, or psychological CLINICAL state
  or history WHEN of a human being WHO and any
  impressions, conclusions, or judgments made
  WHAT regarding that individual within the
  context of the current delivery of health care to
  that individual."
• Access Services Methods / functions that
• 1. Determine what observations are, or may be,
  available
• 2. Let the user to specify subsets of the
  available observations
• 3. Return to the client
• a link to the observation,
• context and meta- information about the
  observation,
• the observation itself.

7
Common Attributes of Observations

• Observations WHAT
• are made on a specific individual, the patient
  WHO
• represent a snap-shot of that individual in time,
  either at a particular time, or over some
  specified interval of time WHEN
• are made, or recorded, by an instrument or a
  health care professional in some clinical
  context and
• are given (either by the patient, the health care
  institution, or society) some degree of
  confidentiality.

8
Observation Objects

• At least two flavors of observation objects have
  been identified
• 1. Thin Objects i.e. the observations by
  themselves, or with minimal context information.
• 2. Complete Objects, containing
• - the observation itself
• - context information (including links to
  supporting observations)
• - methods for common operations.

9
Example

• struct Measurement
• QualifiedCode name
• MeasuredValue value
• QualifiedCode units
• TimeStamp timeStamp
• QualifiedCodeSeq flags

10
Observation Classes and Uses

• Measurements past, current, and future - for
  display, analysis, and alerts.
• Laboratory results for display, analysis, and
  alerts.
• Images for display within the EPR.
• Captured speech for report transcription and
  review, v-mail, etc.
• Reports for editing, review, sign-off, and
  display within the EPR.
• Scanned documents for display within the EPR.

11
Request Filtering

• Simple queries on Who, When, and / or What.
• Who one or more PIDs, or dont-care
• When specific Date-Time, Time-Span, or
  conceptual concept, e.g. Latest, Previous,
• What LQS conceptCode for general or specific
  observation type using HL7 or DICOM or some other
  public and/or private lexicon.

12
Scope

• Explicit support for
• Measurements
• Generic support for
• Documents
• Images
• No support for
• Conversion between data formats
• May perform a subset of the envisioned RLS
  functionality

13
Background of Submitters
14
Usage Examples (1)

• Client queries PIDS to retrieve the Patient Ids
  and their related Id Domains
• Client queries Trader to find the COAS servers
  within those Id Domains
• Client queries each COAS to find out what
  information is available for the patient
• Client accesses (requests) specific observations
  from one or more COAS servers.

15
Usage Examples (2)

• A physician refers a patient for an imaging study
• The physician is notified when the imaging report
  is available.
• The physician
• logs onto a workstation,
• selects the patient's record, and
• requests the report.
• The report refers to visual features seen in the
  images.
• The physician
• requests non-diagnostic quality versions of the
  images, and
• views both the report and one or more images.

16
Usage Examples (3,...)
17
Open Issues

• Information model - we intend to define a model,
  and have several to build on
• Dynamic Discovery - Yes, but what and how?
• Value Domains - we need to clarify the
  relationship between COAS and LQS.
• Componentization - Similar to PIDS and LQS
• Publish/subscribe - Yes, but Event or
  Notification?
• Synchronous vs Asynch- Should we support both?
• Observation Confidentiality - Access control,
  How?