1 of 36

EBM-on-FHIR�(Evidence Based Medicine-on-FHIR)September 24, 2020

Brian S. Alper, MD, MSPH, FAAFP, FAMIA

CEO, Computable Publishing LLC

Project Lead, EBMonFHIR

Project Lead, COVID-19 Knowledge Accelerator (COKA)

Slides available at https://tinyurl.com/COKAfolder

Questions: balper@computablepublishing.com

2 of 36

Introductions

Health Level 7 International (HL7®) is a standards development organization

Fast Healthcare Interoperability Resources (FHIR®) is an HL7 standard that SOLVES the BIG problem in making healthcare data Findable, Accessible, Interoperable and Reusable (FAIR)

Evidence-Based Medicine/Practice (EBM/P) is a community that has the BIG problem of scientific results (evidence) is not FAIR, and there is no universal standard

EBMonFHIR is an HL7 project to extend FHIR to EBM knowledge assets – make evidence FAIR – started May 2018

COVID-19 Knowledge Accelerator (COKA) is a virtual organization (> 25 groups in > 7 countries meeting 12 times/week) to accelerate EBMonFHIR and related efforts for COVID-19 knowledge – started March 2020

3 of 36

The Knowledge Ecosystem – Clinical Knowledge View

Making decisions and doing things.

ACTION

Advice and information to facilitate making decisions.

GUIDANCE

The observations and measurements of things.

DATA

The analysis of data and understanding of likelihoods of things.

EVIDENCE

Outer ring - General Layer

4 of 36

The Knowledge Ecosystem – Clinical Knowledge View

Making decisions and doing things.

ACTION

Advice and information to facilitate making decisions.

GUIDANCE

The observations and measurements of things.

DATA

The analysis of data and understanding of likelihoods of things.

EVIDENCE

Outer ring - General Layer

Specific Layer

5 of 36

The Knowledge Ecosystem – Clinical Knowledge View

Making decisions and doing things.

ACTION

Advice and information to facilitate making decisions.

GUIDANCE

The observations and measurements of things.

DATA

The analysis of data and understanding of likelihoods of things.

EVIDENCE

Outer ring - General Layer

Specific Layer

Computable Layer

6 of 36

The Knowledge Ecosystem – Clinical Knowledge View

Making decisions and doing things.

ACTION

Advice and information to facilitate making decisions.

GUIDANCE

The observations and measurements of things.

DATA

The analysis of data and understanding of likelihoods of things.

EVIDENCE

Outer ring - General Layer

Specific Layer

Computable Layer

Executable Layer

7 of 36

The Knowledge Ecosystem – Clinical Knowledge View

CORE

Making decisions and doing things.

ACTION

Advice and information to facilitate making decisions.

GUIDANCE

The observations and measurements of things.

DATA

The analysis of data and understanding of likelihoods of things.

EVIDENCE

Outer ring - General Layer

Specific Layer

Computable Layer

Executable Layer

8 of 36

What do we need to do in early developmental stages?

  1. Agree to standard format for expression – schema/model of FHIR Resources.
  2. Agree to common terminologies – code systems, values sets.
  3. Develop tools for initial implementation – demonstrating that data exchange can work.

Demonstrating computable data exchange instead of digital data exchange.

  • Adapting tools for early adopter real-world implementation.
  • Develop an Implementation Guide.

Learning from real-world steps BEFORE writing how-to.

9 of 36

1. schema/model of FHIR Resources

FHIR Resources at Maturity Level 1

  1. Evidence
  2. EvidenceVariable
  3. Statistic (Datatype)
  4. OrderedDistribution (Datatype)

FHIR Resources at Maturity Level 0

  1. Citation
  2. EvidenceReport

10 of 36

What does computable evidence look like?

11 of 36

DIGITAL EVIDENCE

Published July 17, 2020

https://www.nejm.org/doi/pdf/10.1056/NEJMoa2021436

12 of 36

DIGITAL EVIDENCE Published July 20, 2020  https://gps.health/coka/reports/EvidenceReport/28

13 of 36

14 of 36

15 of 36

16 of 36

17 of 36

18 of 36

19 of 36

20 of 36

21 of 36

22 of 36

23 of 36

24 of 36

25 of 36

26 of 36

27 of 36

28 of 36

EvidenceVariable Resource

29 of 36

1. schema/model of FHIR Resources

FHIR Resources at Maturity Level 1

  1. Evidence
  2. EvidenceVariable
  3. Statistic (Datatype)
  4. OrderedDistribution (Datatype)

FHIR Resources at Maturity Level 0

  1. Citation
  2. EvidenceReport

30 of 36

2. Code systems and value sets

For these 6 evidence-related FHIR Resources we created 35 different value sets (and thus 35 different code systems) where no existing code systems were adequate.

The process (derived from FHIR Resource generation) was not set up for creation of code systems.

We created a Code System Development Protocol

1. Assemble an expert working group.

2. Identify tools or systems commonly used today to express these concepts.

3. Map out a single list of non-redundant concepts for common needs.

4. Map related terms and definitions in all freely available ontologies.

5. Determine the optimal phrasing for terms and definitions in the code system.

6. Achieve (near-)universal agreement in the code system.

7. Evaluate implementation of the code system.

8. Maintain systems for continuous code system refinement.

31 of 36

2. Code systems and value sets

We are working now on creating 4 Code Systems:

Study Design Code System to describe methodology characteristics of scientific observations such as randomized assignment, crossover allocation to cohorts, and qualitative analysis.

Statistic Type Code System to describe the statistics reported such as mean, median, proportion, relative risk, absolute risk difference, p value, confidence interval, and I-squared measure of heterogeneity.

Statistic Model Code System to describe the model characteristics such as linear regression, random-effects analysis, and tau estimation method.

Risk Of Bias Code System to describe the concerns with methods or reporting of scientific observations such as comparator selection bias, gaps in blinding, and selective outcome reporting.

 

32 of 36

33 of 36

Early mapping work led us to realize we needed more structure for how we create the code systems

34 of 36

2. Code systems and value sets

You can join any of the Expert Working Groups.

Read the Invitation to join Expert Working Group

 

Complete the Code System Development Intake Form

36 people from 19 countries (5 continents) have signed up so far

35 of 36

3. Demonstrating that data exchange can work

Content Citation and Classification Tools Development Work Group

  • Completed mapping between MEDLINE (XML, txt, RIS) elements, CORD-19 JSON elements, and FHIR Citation elements
  • Group goal = Create infrastructure standards and tools to support interoperable, reusable efforts for Citation Repositories

Evidence Evaluation and Reporting Tools Development Work Group

  • Simple data exchange demonstration underway
    • Systematic Review Data Repository (SRDR) sends outcome definition data as FHIR EvidenceVariable Resource in JSON
    • Computable Publishing displays human-readable view in Outcome Definition Viewer
  • Project in discussion to develop tools for data entry for clinical trial reports

Systematic Meta-Review Project Group

  • Facilitate the production of systematic reviews of systematic reviews with expression in FHIR format.
  • Start with a systematic meta-review of steroids as treatment for COVID-19.

 

36 of 36

COKA Weekly Meeting Schedule

Day

Time (Eastern)

Team

Monday 

7-8 am 

Project Management

Monday 

9-10 am

Systematic Meta-Review

Monday

4-5 pm

Statistic Model Code System Development

Tuesday 

10-11 am 

Evidence Evaluation and Reporting Tools Development

Tuesday 

2-3 pm 

Study Design Code System Development

Wednesday

8-9 am 

Knowledge Ecosystem Liaison

Wednesday

9-10 am

Statistic Type Code System Development

Thursday 

7-8 am 

Project Management

Thursday

9-10 am

Content Citation and Classification Tools Development

Thursday 

4-5 pm 

Project Management

Friday

9-10 am 

Risk of Bias Code System Development

Friday

10-11 am 

Communications (Awareness, Scholarly Publications)

For all of these meetings: Join Microsoft Teams Meeting

If you wish to sign up or learn more about COKA please go to

https://www.gps.health/covid19_knowledge_accelerator.html

Questions: balper@computablepublishing.com