InferMed

1
(No Transcript)
2
SNOMED CT and Computer-interpretable Guidelines

• Dr Robert Dunlop
• Clinical Director
• InferMed Ltd

3
Presentation Overview

• Who is InferMed?
• What is AREZZO?
• What are Computer-interpretable Guidelines
  (CIGS)?
• How has SNOMED CT helped?

4
Who is InferMed?
Privately held UK company 95 of revenue from
outside UK 23 staff
5
What is AREZZO?

• AREZZO is a software technology for
• Creating guidelines, protocols, care pathways and
  other decision support applications
• Running guidelines and protocols at the
  point-of-care for individual patients
• Can link to electronic medical records

6
AREZZO Technology

• Developed over 10 years in the Advanced
  Computation Laboratory of CRUK
• Based on a sound theoretical framework developed
  by Professor John Fox
• Using argumentation approach, derived from
  observation of expert decision-making
• Several trials published
• CAPSULE, RAGS, LISA, HAVANNA, and Breast Cancer
  Triple Assessment

7
AREZZO Guidelines

• AREZZO facilitates active protocols and
  guidelines
• Most guidelines comprise text or html documents
  that may be very lengthy
• Arezzo customizes guidelines to the individual
  patient
• Clinician only interacts with relevant parts of
  guideline

8
AREZZO Building Blocks

• AREZZO guidelines are made up of
• Tasks and collections of tasks
• Data Items
• Tasks and data items have attributes
• Define the behaviour of the guideline during
  enactment by AREZZO Performer

9
AREZZO Overview
10
AREZZO CIGs and SNOMED CT

• HS Copenhagen District Hospitals EMR Project
• COCOON Project
• MAPS Project
• CREDO Project

11
HS EMR Project

• Copenhagen District Hospitals and 2 Danish Health
  Regions
• AREZZO selected for CIGS
• Guideline content is SNOMED encoded
• AREZZO Composer is being linked to SNOMED browser
• HL7 Code attributes added to tasks and data items

12
HS EMR Project

• All data collected and all tasks executed during
  guideline enactment are automatically SNOMED
  coded
• Data collection and coding is part of the process
  of care
• Current status
• Field trials commencing in April 2006

13
COCOON Project

• EU-funded 6th Framework healthcare IT project
• Deliver CIGs (and other services) to GPs in EU
  Health Regions
• Integrated with EMR systems
• Currently Lombardy, Epirus and Brussels Regions

14
COCOON Project

• Multiple guidelines, including Parkinsons
  disease
• Subset of relevant SNOMED CT hierarchy placed in
  AREZZO Class Editor
• Drugs and Parkinsons related diagnoses
• Signed off by clinical experts in project

15
COCOON Project

• At run-time, patient medications and diagnoses
  are mapped to SNOMED CT from EMR
• Eg from ICD and ATC codes in Lombardy Region
• Codes are passed to COCOON decision support via
  HL7

16
COCOON Project

• SNOMED codes are checked against the Class Editor
• Logical inference is used to determine the class
  of drug and/or type of movement disorder
• Some Task state triggers based on this inference
• Eg data on side-effects requested according to
  the parent drug class.

17
COCOON Project

• Current status
• Field trials have commenced in Lombardy
• About to commence in Epirus and Brussels
• Project to be extended to health regions in new
  Member states

18
MAPS Project

• National Institute Aging funded project
• Pain management guidelines in US nursing homes
• Linked to the Minimum Dataset
• Supported by Centers for Medicare and Medicaid
  Services (CMS)
• Collaboration with Brown University

19
Data Collection on PDA Device
20
MAPS Faxback Report to GPs
21
MAPS Project

• Current status
• Evaluation underway in Rhode Island nursing homes

22
CREDO Project

• CRUK trial to improve consistency, quality and
  safety in care of cancer patients
• Suite of CIGs for breast cancer
• Screening
• Genetic risk assessment
• Early referral
• Triple assessment

23
CREDO Project

• Palliative care CIG is being developed
• Pain and other symptom management
• Ontology of SNOMED terms
• Pathology of breast cancer
• Patterns of spread
• Pain syndromes

24
CREDO Project

• SNOMED CT codes being used to support logical
  inference about
• Cause of pain syndrome
• Visceral vs somatic vs neuropathic
• Cancer-related or not
• Recommended investigations and treatments

25
Lessons

• CIGs provide a valuable method of collecting and
  coding high-quality data as part of the process
  of care
• Drug-based logical inference can be readily
  supported from SNOMED CT

26
Lessons

• More sophisticated CIG applications require
  significant extra work
• Limited semantics
• Limitations in the breadth of terms
• Only 40 of concepts available in breast cancer
  domain for the CREDO Project
• Clinician sign-off needed for knowledge base
• Particularly for logical inference