- Mayo Clinic Informatics
Download
Report
Transcript - Mayo Clinic Informatics
NHS Modelling Efforts –
ISO13606 adoption and beyond
Dr. Rahil Qamar Siddiqui
Health and Social Care Information
Centre,
NHS, England
Background: Information Standards
use in UK
• The Health and Social Care Act 2012 states that the
following must have regard to an Information
Standard published under the Act:
Secretary of State for Health
NHS England
Public bodies involved with health services or adult
social care
Anyone providing publicly funded health services or
adult social care commissioned by or on behalf of a
public body
Information Standards development
Source: www.hscic.gov.uk
Implementing Standards
• The Information Standards Delivery department
provides the building blocks to enable development
and implementation of information standards
Terminologies
Classifications
• Read
• SNOMED CT
• dm+d
• OPCS-4
• ICD-10
NHS Data
Dictionary
• Data
Collections
• Data Sets
NHS Data
Dictionary for
Care ?
• Data recording/
record keeping
• Terminology
Binding
HSCIC.GOV.UK -> Information Standards
HSCIC.GOV.UK .. With DD4C
The DD4C Process ..(1)
The DD4C Process ..(2)
Phase 1
Existing Content
Catalogue
• SNOMED CT Subsets
• Data Collections
•NHS Messaging Specs
•Royal Colleges Headings
•Define metadata of all
content
•Metadata model based
on ISO 13119
Publish
Harmonise
• ISO13606-based logical
models for Royal Colleges
headings
• ISO13606-based logical
models for clinical record
keeping
Publish
Catalogue Metadata
Logical Models for Care ..(1)
• The aim
Provide clinically assured, quality assessed, process-driven logical
representations of health care records
Single reference point for all product-dependent modelling work
such as NHS Data Dictionary and NHS Messaging Specifications
Provide metadata for our logical models to provide valuable
information about the models as well as to allow associations with
other content such as SNOMED CT subsets, message templates etc
Allow multiple format download of our logical models as a free
public resource: ADL, UML, HTML, XML, JSON, Mindmap, Word etc.
Logical Models for Care ..(2)
ISO 13606 Extract Reference Model adopted
Revision proposed to ISO 13606 Demographics Model
(awaiting adoption)
SNOMED CT used as the reference terminology for binding to
clinical content
Demographic and participant-related vocabulary for other
content (to be decided)
LinkEHR Editor used for logical modelling (pending approval)
Logical Modelling Guideline document to help with consistent
modelling and terminology binding (draft status)
Logical Models for Care ..(3)
Two level modelling process
• High level models (for initial discussion with health informaticians
and clinicians)
• Logical models (technical models for discussion with health
informaticians and technical modellers)
Transformation process (initially manual but possible
automated transformation in future?)
• NHS Data Dictionary models
• NHS Messaging Specifications (CDA, FHIR?)
• Other consumers (NHS as well as external) that require information
models
Logical Models for Care ..(4)
• ISO13606-based models
for Royal College headings
• ISO13606-based models
for record keeping
SNOMED CT
CIMI Models
International
External Reference Resources
• Discharge Summary
• Outpatient
• Referrals
• Diagnosis
• Blood Pressure
• Medications ....
SemanticHealthNet
patterns
CONTSys
Logical Record
SNOMED CT
Architecture
UK Extensions
(LRA)
Internal Reference Resources
Example: Royal Colleges Headings
Level one: High level (discussion)
models
Level one: High level Examination
Finding
Level two: Logical models
Bottom-up modelling approach ..(1)
CLINICAL STATEMENT SUB-PARTS: CLUSTER & ELEMENT
• Where clinical statements have a more complex structure, each of
the sub-parts of a clinical statement should be modelled as
independent, reusable models using the ITEM classes: CLUSTER and
ELEMENT.
• These clinical statement sub-parts may represent both the core
data describing parts of a statement and, optionally, other details
the
– examination method (technique and/or device used),
– the patient’s physical state,
– assertion status such as presence/absence, normality/abnormality
indicators of the core value.
• When modelling a clinical statement sub-part it is appropriate to
start at the CLUSTER class as the ISO 13606 Reference Model
allows for nesting of CLUSTERs within CLUSTERs as well as inclusion
of one or more ELEMENTs which hold the actual data
CLUSTER: Blood Pressure
Measurement
CLUSTER: Blood Pressure
Measurement State
Level two: Logical models
Bottom-up modelling approach ..(2)
CLINICAL STATEMENT: ENTRY
• Clinical Statements in this document refer to clinical
observations, requests, results, actions, plans, instructions,
intentions, interpretations, reasoning etc. which may have
a simple or a more complex structure.
• Each of the CLUSTERs represents a single, indivisible, and
reusable clinical statement sub-part which are then
collectively modelled in an ENTRY, which is used to record
information in an EHR as a single, indivisible clinical
statement. The ENTRY class also helps to provide the set of
context attributes that help to facilitate the safe
interpretation of the clinical statement.
ENTRY: Blood Pressure
Level two: Logical models
Bottom-up modelling approach ..(3)
COLLECTIONS: SECTION
• The term Collections is used to refer to a group of clinical
statements which might be viewed together under a
common heading for human consumption. These
collections are modelled using the CONTENT class:
SECTION.
• Within DD4C, the SECTION class should be used only when
implementation-specific requirements warrant the need for
headings such as is the case with the Royal Colleges
headings.
• NOTE: Else the CLUSTER and ENTRY models should be used
to logically model the clinical content irrespective of how it
is desired to be viewed by specific user communities.
SECTION: Vital Signs
SECTION: Examination Findings
Level two: Logical models
Bottom-up modelling approach ..(4)
ORGANISERS: COMPOSITIONs and FOLDERs
• Collections of clinical statements could be included in highlevel organiser groups to which authoring, attesting, and
auditing information can be associated. These organisers
help in human readability, storage, and communication
rather than machine computability and semantics.
• COMPOSITIONs are typically used to communicate extracts
of records both internally within an organisation such as
Admission and Handover records, as well as records that
might be communicated externally such as Discharge,
Outpatients, and Referral records. Each instance of a
COMPOSITION is about a single subject of care.
COMPOSITION:
Outpatients
Logical Models for Care .. (5)
• The project started mid-November 2013
• First batch of logical models to be produced by end of
April 2014
– Royal College Headings
• 2-3 detailed headings e.g. Vital Signs, Diagnosis, Procedures
– Medications (which includes Dose Syntax)
• Dose Syntax, LRA models, other NHS work, Reviewing Scottish work
on Medications (using openEHR archetypes)
• Terminology Binding strategies underway. All ‘clinical’
logical models to be bound to SNOMED CT
appropriately.
• ‘Demographic’ logical models along with participant
information to be included as per requirements.
Logical Models for Care ..(6)
• Future Aspirations!
Large, open library of logical models for primary consumption within NHS
and its providers
Provide implementable NICE Guidelines : supported by logical models
Provide a valuable resource to Standardisation Committee for Care
Information (SCCI) (old ISB) to harmonise the content within the
standards approved by them to have a better view of data and
information flow across existing SCCI standards
Provide rules for applying a combination of logical models along with the
implementable NICE Guidelines and (potentially) SCCI standards
mandated to arrive at decision support logic.
NICE - National Institute for Health and Clinical Excellence
Thank you
Questions?
[email protected]