Amendment History

Table of Contents
1 Glossary
1.1 Domain Terms
2 Introduction
2.1 Purpose
2.2 Audience and stakeholder domain
2.2.1 Input from stakeholders
3 Solution Development
3.1 Initial Design
3.1.1 Outline of initial design
3.1.2 Create grouper concepts for finding site
3.1.3 Create grouper concepts for etiology
3.1.4 Concepts edited
3.1.5 Evaluation of Design
3.2 Iteration One
4 Quality program criteria
4.1 Quality metrics
4.1.1 Quality metric 1
4.1.2 Quality metric 2
4.2 Use case scenarios
4.2.1 Finding site use case
4.2.2 Etiology use case
4.3 Test cases
5 Updated Project Resource Estimates
5.1 Scope of construction phase
5.2 Projection of remaining overall project resource requirements
5.2.1 Expected project resource requirement category
5.2.2 Expected project impact and benefit
5.2.3 Indicative resource estimates for construction, transition and maintenance:

Glossary

Domain Terms

Introduction

Purpose

The purpose of this project is to consider the reorganization of descendants of Acute Inflammatory Disease (disorder), by site and etiology

Audience and stakeholder domain

The audience for this document includes all standards terminology leaders, implementers and users but is especially targeted at those with an interest in reorganizing Acute inflammatory disease (disorder) concepts by site and etiology as opposed to the current hierarchical structure.
A further significant audience is the community of SNOMED authors that may be requested to implement the recommended specification.

Input from stakeholders

The request is to reorganize the descendants of Acute inflammatory disease (disorder) by site and etiology and work to eliminate any stated descendants of Acute inflammatory disease (disorder) so that the taxonomy will be automatically classified based on the definition of each concept. As part of the elaboration phase, use of acute timing and acute morphologic change should be explored as well.

Solution Development

Initial Design

Outline of initial design

Create grouper concepts for finding site

Finding site grouper concepts should be created that reflect the finding site group concepts used in other taxonomies. For example, the Disorder by body site (disorder) taxonomy will be examined, but only the sites that will actually inherit an Acute inflammatory disease (disorder), will be added.

Create grouper concepts for etiology

Causative Agent grouper concepts should be created that reflect the Causing Agent group concepts used in other taxonomies. Unfortunately, there is not a simple example to point to like there is with the Disorder by body site. In some cases, the concepts are there, but just more integrated (Infections disease) rather than explicitly called out (Infection caused by an organism) As part of elaboration, several proposals for the etiology grouper concepts should be considered.

Concepts edited

4 concepts were revised to eliminate stated children of Acute inflammatory disease (disorder)
19272000Acute cervicitis
302911003Acute lingual tonsillitis
237037006Acute pelvic inflammatory disease
190293001Acute thyroiditis
After this change, and classification, 70 immediate inferred children of Acute inflammatory disease (disorder) remain. 68863007 Acute syphilitic meningitis (disorder) is a representative example of an inferred concept that is a direct inferred descendent of Acute inflammatory disease (disorder).
Added a new concept to pull all inflammatory diseases caused by infectious process as follows:

This concept properly classified underneath Acute inflammatory disease (disorder), resulting in 63 inferred children. Acute inflammatory disease (disorder) now only has 55 children, some of which should be under Acute inflammatory disease caused by infectious process (disorder) such as Acute infective balanitis (disorder), indicating incomplete modeling of these concepts.
(missing causative agent of organism).

Adding concept for body site:

Classification after adding this concept resulted in the following inferred taxonomy:

Evaluation of Design

Criterion 1: Acute inflammatory disease should have NO stated children as any concept that includes both an inflammatory morphology and a clinical course of "Sudden onset and/or short duration" or its children would automatically classify under this high level grouper concept.
Criterion 2: Immediate inferred children of Acute inflammatory disease should have only inferred children that represent acute inflammation organized by organ system, or acute inflammation organized by etiology.

Exceptions and Problems

I noted that there where two concepts in morphology that looked very much the same, and might be used as synonyms even though they where two distinct concepts: 
SNOMED integer id    4532008          Acute inflammation (morphologic abnormality)
SNOMED integer id    409776007      Acute inflammatory morphology (morphologic abnormality)
Where Acute inflammatory morphology is-a Acute inflammation
Acute inflammatory morphology is used to define: 

I thought that by changing "Associated morphology: Acute inflammatory morphology" to "Associated morphology: Acute inflammation" I might get more children to acute inflammatory disease.
Before the change, there where 498 "kind-of" concepts, and after the change & running the classifier there where only 483 "kind-of" concepts.  I expected the number to go up, since I am generalizing the criterion for meeting the criterion...
One example of concept that is no longer a "kind-of" after the change is Acute abscess [of] lymph node (note that the definition of this concept was not changed): 

Which uses associated morphology Acute Abscess, which is a primitive subtype of Acute inflammatory morphology. 

So changing the definition of Acute Inflammatory Disease to: 

And then classifying, the inferred taxonomy for Acute abscess [of] lymph node is: 
(note that Acute Inflammatory Disease is not one of the parents). 

I would expect that Acute Inflammatory Disease would be one of the parents since Acute Abscess -> Acute inflammatory morphology -> Acute inflammation… Is my expectation wrong? 
I suspect that this may have something to do with Acute Abscess being a primitive child of Acute inflammatory morphology, and maybe the classification algorithm is overlooking something? I placed an inquiry to CSIRO on this topic.

Response to Review

The reviewer requested to know what would happen if we retired "Acute inflammatory morphology" and moved the four child concepts other than "Acute inflammation" under 45959006 | Acute suppurative inflammation (morphologic abnormality) |.
I reconstructed this example with the 20150731 version of SNOMED previously used, and made the requested changes. The four children of "Acute inflammatory morphology" other than "Acute inflammation" where moved under "Acute suppurative inflammation (morphologic abnormality)."
The is-a relationship from "Acute inflammation" to "Acute inflammatory morphology" was retired.
In addition to the changes requested, I had to change the "associated morphology" restriction from "acute inflammatory morphology" to "acute inflammation" for the following concepts:

• Acute inflammatory disease caused by bacterial infection

• Acute pelvic inflammatory disease

• Acute inflammatory disease caused by infectious process

• Acute uterine inflammatory disease

• Acute inflammatory disease affecting body site

• Pyosalpinx

After these changes, "Acute inflammatory morphology" had no children and had no role relationships pointing to it, and it was retired.
After classification, and recompilation of the Acute Inflammatory disease children (refset) there are now 501 members of this refset, and increase over the previous 483 members computed prior to the retirement of the "Acute inflammatory morphology" concept. This increase indicates to me that the taxonomy is improved, and the sensitivity of recall is improved. And "Acute inflammatory disease" is now one of the parents of "Acute abscess [of] lymph node".
The resulting inferred taxonomy is shown below.

Design Strengths

It is relatively simple, and consistent with the organization in the reset of SNOMED.
Criterion 1: Acute inflammatory disease should have NO stated children was satisfied.
Criterion 2: Immediate inferred children of Acute inflammatory disease should have only inferred children was satisfied. .

Design Weakness

None noted.

Design Risks

Iteration One

The initial iteration was sufficient.

Quality program criteria

Quality metrics

Quality metric 1

 

Quality metric 2

 

Use case scenarios

Below are use case scenarios that serve at manual test cases that can be performed by classification and inferred taxonomy navigation.

Finding site use case

A user should be able to navigate to inflammatory conditions by navigating to the site of that condition. The navigational concepts should be kept to a pragmatic limited set of navigational concepts, and should not be the transitive closure of all children concepts.

Expected Setting

Taxonomy navigation

Data capture requirement

When capturing data from selection accomplished by navigating the taxonomy.

Data retrieval requirement

When retrieving data from concepts specified by class (class based query)

Etiology use case

A user should be able to navigate to inflammatory conditions by navigating by the known etiology of that condition.

Expected Setting

Taxonomy navigation

Data capture requirement

When capturing data from selection accomplished by navigating the taxonomy.

Data retrieval requirement

When retrieving data from concepts specified by class (class based query)

Test cases

The use cases above have all been accomplished as manual test cases as demonstrated by classifying the resulting definitions, and navigating the inferred taxonomy to show that finding site and etiology use cases are satisfied.

Updated Project Resource Estimates

Estimated project size is small, it is essentially already complete. I estimate less than an additional 40 hours to include additional quality assurance and secondary review of my changes. This document has sufficient detail to reproduce all of my changes.

Scope of construction phase

Limited to a small taxonomy of acute inflammatory diseases. Skills required are a clinician with reasonable familiarity is inflammatory pathology and etiology. Expertise in histopathology or infectious diseases is not required.

Projection of remaining overall project resource requirements

Expected project resource requirement category

It is "fast track" and required minimal project management.

Expected project impact and benefit

Updated view of impact and benefit, organized by stage if the project is to be staged

Indicative resource estimates for construction, transition and maintenance:

Construction and transition phase: ~10 new concepts to be authored
Maintenance phase: No expected additional concepts are required.

---