Medical Coding

1 Algorithm Details
- 1.1 MedDRA
  - 1.1.1 Example with full match:
  - 1.1.2 Example with partial match:
- 1.2 WHO Drug

Role: Medical Coder

Provided data fields have been configured for medical coding via MedDRA and WHO Drug dictionaries, data collected against those fields becomes eligible for coding in the Medical Coding > Data module. Here, ClinSpark provides an opportunity to auto-code terms by looking for a best match in the respective dictionary or the opportunity to manually code each term by browsing the dictionary tree. If collected data is updated after initial coding has occurred, the user is prompted to re-code the updated term.

Algorithm Details

ClinSpark has a built-in feature designed to streamline “Medical Term” coding. This feature allow an employee to find a matching term in the corresponding Medical Dictionary automatically. In some dictionaries medical terms can be located in different layers (example would be different ATC levels) and in different paths (example would be different drug names designations, i.e. trade name vs generic name). This article explain how auto-coding works and the algorithm behind confidence level calculations.

MedDRA

For this dictionary we only search against LLT, but we perform search in three different ways relaxing search contains more and more and thus lowering confidence level.

Confidence	Search Target	Search Match	Example

Confidence	Search Target	Search Match	Example
50%	LLT	Partial Match on first part of the term	Influenza B → Influenza
75%	LLT	Partial Match on entire term	Influen → Influenza
100%	LLT	Full Match on entire term	Influenza → Influenza

LLT* = Lowest Level Term

Example with full match:

Example with partial match:

WHO Drug

For this dictionary we do not calculate confidence level, and instead we allow an employee to select best option from the drop down list of matching items. We use the following sequence of searches to build a list of matching items:

Priority	Search Target	Search Match	Example

Priority	Search Target	Search Match	Example
1	Any Drug Name (Generics and Non Generics)	Full Match	Remicade → Remicade
2	Any Drug Name (Generics and Non Generics)	Partial Match	Remi → Remicade, Infliximab

Example with full match in Generic and Trade names:

Example of multiple matches of the same drug bound to a different placers with in ATC ontology:

Currently, when adding a Medication or a Medical Term, the auto-coding feature allows for subsequent edition of the hierarchical tree. This may be undesirable, as it can lead to false classification. In a later release of ClinSpark, the feature will likely restrict the edition of the hierarchical tree to avoid this challenge.

When adding a Medication, it is possible to change the hierarchical tree, from top to bottom. Changes do not affect downstream fields. The original term is non-editable. All changes (including original coding) remain in the audit trail.

When adding a new Medical Term, the top to bottom changes for the hierarchical tree are allowed, with the request to update the fields downstream. The original term is non-editable. All changes (including original coding) remain in the audit trail.