Algorithm Descriptions

Algorithms are a named set of instructions used in the Symedical mapping process that determine source to target map candidates. The following algorithms are included in Symedical:

Please Note: The algorithm descriptions in this section include examples of the results that may be returned. The examples are provided for illustration purposes only and do not represent all results that may be returned.

Characteristics

Includes a set of "characteristic" algorithms and a set of weighted characteristic algorithms.

Use the All, Exact, and Primary characteristic algorithms to apply those algorithms to the mapping process of each domain contained within a map’s composite source catalog. These domain-specific algorithms are particularly useful when creating map templates.

Use the weighted characteristic algorithms to enhance the number of relevant candidates returned based on assigned importance.

Default criteria are set for a given domain by Clinical Architecture within the Domain Manager Algorithm Weight tab. If you wish to define your own defaults for new maps to inherit, it is best practice to do so within the Map Template Algorithms tab. Alternatively, define weights for specific maps within Map Manager Algorithms.

Characteristic: All

An exact word match where all components of the source description are found in the target description. Using a drug domain example, the following candidate will be returned because all populated domain characteristic values of the source term match those of the target term.

Source Description

Target Description

Abacavir Sulfate 300 mg Tablet

Abacavir Sulfate 300 mg Oral Tablet

Characteristic: Broader Target

Compares secondary characteristic values to return candidates based on the:

  • Broadness of the target term where its secondary characteristics either match the source term or are blank.

  • Algorithm configuration to exclude specific secondary characteristics that are not clinically relevant when determining candidates.

  • Algorithm configuration to rank the results by defining clinically relevant characteristic sets in descending order.

By default, the algorithm determines rank based on the number of matching characteristic values. These settings can be modified at the domain (default), map template, and map levels within the following Symedical applications:

  • Domain Manager Algorithm Settings

  • Map Template Algorithms

  • Map Manager Algorithms

The Broader Target algorithm returns all viable matches as candidates and maps only when one mappable candidate is returned.

Characteristic: Exact

An exact word match where all components of the source description match all of the components of the target description without anything extra in the target term. Using a drug domain example, the following candidate will be returned because all components match.

Source Description

Target Description

APAP 325 mg Tab

Acetaminophen 325 milligram tablet

Characteristic: Primary

An exact word match on the primary characteristic. Using a drug domain example, the following descriptions match on the ingredient primary characteristic.

Source Description

Target Description

furosemide miscellaneous powder (GM)

furosemide 10 mg/ml Oral Solution

Partial Token Weighted Overlap Similarity (PWOS)

For source and target terms with partial primary characteristic match (matching primary characteristic token(s) of specified number and length), PWOS takes into account the designated importance (Critical, High, Medium, Low, None) of secondary characteristics and generates a similarity score. The similarity score takes into account primary and secondary characteristic percent overlap.

Algorithm Requirements: Map source terms are not processed by these algorithms during map analysis if their associated domain has the Default is Null check box selected and a default Primary Characteristic value specified in the Domain Manager Characteristics Tab.

Weighted Overlap Similarity (WOS)

For source and target terms with matching primary characteristics, WOS takes into account the designated importance (Critical, High, Medium, Low, None) of secondary characteristics and generates a similarity score. The similarity score is the percent overlap of source and target characteristic weights.

Algorithm Requirements: Map source terms are not processed by these algorithms during map analysis if their associated domain has the Default is Null check box selected and a default Primary Characteristic value specified in the Domain Manager Characteristics Tab.

Domain-Specific Algorithms

Domain-specific algorithms are available according to the domains assigned to the map's source catalog. A domain-specific algorithm returns candidates in which the primary and all specified secondary characteristics match.

Drug and Lab Test domains are the most commonly used domains in Symedical. See the following for algorithms that are specific to the Drug and Lab Test domains:

Algorithm Requirements: Source terms are not processed by these algorithms during map analysis if their associated domain has the Default is Null check box selected and a default Primary Characteristic value specified in the Domain Manager Characteristics Tab.

Tip: If the map or map template contains a Composite domain, you can use the Domain Independent algorithms (Characteristic: All, Exact, and Primary) rather than adding the All, Exact, and Primary algorithms for each available domain.

Drug

This topic defines the Drug domain algorithms.

Exact

An exact word match, where all components of the source description match all of the components of the target description without anything extra in the target terms.

Source Description

Target Description

Abacavir Sulfate 300 mg Oral Tablet

Abacavir Sulfate 300 mg Oral Tablet

All

An exact word match where all the components of the source description are found in the target description. The following candidate will be returned because all populated domain characteristic values of the source term match those of the target term.

Source Description

Target Description

Abacavir Sulfate 300 mg Tablet

Abacavir Sulfate 300 mg Oral Tablet

Ingredient, Strength, Dose Form, Route

An exact word match on ingredient, strength, dose form, and route.

Source Description

Target Description

DOPamine 40 mg/mL intravenous solution

Dopamine HCL 40 mg/mL Intravenous Ampul

Ingredient, Strength, Dose Form

An exact word match on ingredient, strength, and dose form.

Source Description

Target Description

Warfarin 10mg Capsule

Warfarin 10mg Cap

Ingredient, Strength, Route

An exact word match on ingredient, strength, and route.

Source Description

Target Description

albuterol 4 mg oral tablet, extended release

Albuterol Sulfate 4 MG Oral Tablet

Ingredient, Strength

An exact word match on ingredient and strength.

Source Description

Target Description

Ampicillin 500 mg Oral Capsule

Ampicillin Sodium 500 mg Injection Vial

Ingredient, Dose Form, Route

An exact word match on ingredient, dose form, and route.

Source Description

Target Description

vitamin a topical cream (grams)

vitamin a 1 unit/mg topical cream

Ingredient, Dose Form

An exact word match on ingredient and dose form.

Source Description

Target Description

zinc oxide topical paste (GM)

zinc oxide 0.2 mg/mg paste

Ingredient, Route

An exact word match on ingredient and route.

Source Description

Target Description

zinc oxide topical paste (GM)

zinc oxide 0.108 mg/mg topical powder

Ingredient

An exact word match on ingredient only.

Source Description

Target Description

furosemide miscellaneous powder (GM)

furosemide 10 mg/ml Oral Solution

Lab Test

LOINC Probability

Used with a LOINC target catalog. A word match made through the Symedical LOINC probability data built from the Regenstrief LOINC Mapping Assistant (RELMA) data and customized weight settings.

Exact

An exact word match where all the components of the source description match all of the components of the target description without anything extra in the target term.

Source Description

Target Description

Albumin MCnc Urine Detection limit <= 20 mg/L QN GC/MS 24H

Albumin MCnc Urine Detection limit <= 20 mg/L QUANT GC/MS 24H

All

An exact word match where all the components of the source description are found in the target description.

Source Description

Target Description

Calcium MCnc Urine

Calcium MCnc 24H Urine

Comp, Spec, RU, Method

An exact word match on the component, specimen type, result unit, and method.

Source Description

Target Description

Methamphetamine Scr Urine Pt (with the result unit of “negative”)

Methamphetamine ACnc Urine Screen (with the result unit of "negative")

Comp, Spec, RU

An exact word match of component, specimen type, and result unit.

Source Description

Target Description

Triglyceride body fluid (with result unit "MG/DL")

Triglyceride Body fld (with result unit "MG/DL")

Comp, RU, Method

An exact word match on the component, result unit, and method.

Source Description

Target Description

Drug Scr Comment Ser Pt (with result unit "exist")

Drug Screen Comment (with result unit "exist")

Comp, RU

An exact word match on component and result unit.

Source Description

Target Description

Creatinine Amniotic Fluid Pt (with result unit "IU")

Creatinine Body Fluid Body fld Pt (with result unit "IU")

Comp, Spec, Prop, Method

An exact word match on component, specimen type, property, and method.

Source Description

Target Description

Ammonia Level Plas Pt Manual Count

Ammonia MCNC PLAS Manual Count

Comp, Spec, Prop

An exact word match on component, specimen type, and property.

Source Description

Target Description

Creatinine MCNC PT PLR FLD ORD

Creatinine MCnc Pt Plr fld Qn

Comp, Prop, Method

An exact word match on component, property, and method.

Source Description

Target Description

Piperacillin Level Ser Pt MIC

Piperacillin level Isolate MIC

Comp, Spec

An exact word match on component and specimen type.

Source Description

Target Description

AMYLASE URINE

AMYLASE ACNC 24 H URINE

Comp, Prop

An exact word match on the component and property.

Source Description

Target Description

Parasites Find urine

Parasites Find Bld

Comp

An exact word match on component.

Source Description

Target Description

PROSTATE SPECIFIC AG NC

PROSTATE SPECIFIC AG MCNC BODY FLD

Lexical

Review the following terms before referencing the section below:

  • Stemming — Normalizes word endings, such as "s," "ing," "ed," etc.

  • Tokenization — Alphabetizes the word order of a term.

The following lexical algorithms are "domain restricted", meaning that they restrict composite catalog mappings to target terms in the same domain when the Domain Independent check box is unchecked.

Please Note: The Referential, Reference Map, and Domain Independent algorithms, as well as the remaining lexical algorithms, never restrict.

Below are the definitions for the lexical algorithms. The Domain Restricted algorithm definitions are found under their respective titles. For example, the Alias to Alias (Domain Restricted) definition is found under Alias to Alias.

Alias to Alias

An exact word match on a source term alias to a target term alias for the same language.

Alias to Alias DiGram

A string matching algorithm that ignores word order and slight misspelling by comparing tokens of 2-character length in source and target term aliases for the same language. If an exact term match is found, only the candidate is returned.

Alias to Alias TriGram

A string matching algorithm that ignores word order and slight misspelling by comparing tokens of 3-character length in source and target term aliases for the same language. If an exact term match is found, only the candidate is returned.

Alias to Lexical

An exact word match on a source term alias to a target term.

Alias to Lexical DiGram

A string matching algorithm that compares tokens of 2-character length in a source term alias to a target term description.

Alias to Lexical TriGram

A string matching algorithm that compares tokens of 3-character length in a source term alias to a target term description.

Digram

A string matching algorithm that compares tokens of 2-character length in source and target term descriptions. The spaces before and after the term description are taken into account and help broaden search results.

Domain Lexical

An exact word and text match after domain-specific transformations.

Please Note: No secondary characteristics are transformed. For example:

Source Description

Target Description

delbase compounding ointment

DELBASE OINTMENT (in which 'compounding' is a trashonym)

Domain Lexical Stem

An exact word and text match after domain-specific and stemming transformations.

Source Description

Target Description

chronic illness

chronically ill

Domain Tokenized

A word match after domain-specific transformations and after the term is alphabetized.

Source Description

Target Description

aspirin/acetaminophen/caffeine

APAP/aspirin/caffeine

Domain Tokenized Stem

A word match after domain-specific transformation, stemming transformation, and the term is alphabetized. For example:

Source Description

Target Description

Visiting Home Dialysis

Dialysis Home Visit

Included Substring

A word match anywhere within the description in which the target is contained within the source.

Source Description

Target Description

Depend Pant Misc

Pant

Lexical

An exact description match.

Source Description

Target Description

aspirin

aspirin

Lexical Digram

A string matching algorithm that compares tokens of 2-character length in source and target term parsed lexical descriptions.

Lexical to Alias

An exact word match on a source term to a target alias.

Source Description

Target Alias

Target Description

CHF

CHF

CONGESTIVE HEART FAILURE

Lexical to Alias DiGram

A string matching algorithm that compares tokens of 2-character length in source term descriptions to a target alias.

Lexical to Alias Included Substring

A word match in which a target alias is contained anywhere within the source term description.

Source Description

Target Alias

Nocturnal Hemodialysis

Hemodialysis

Lexical to Alias Reverse Included Substring

A word match in which the source term is contained anywhere within the target alias description.

Source Description

Target Alias

Acetylcarnitine

High Purity Acetylcarnitine HCL

Lexical to Alias Reverse Substring

An exact word match on a source catalog term to a target catalog alias in which the target alias description begins with the source description.

Source Description

Target Alias

Vertigo

Vertigo mild

Lexical to Alias Substring

An exact word match on a source catalog term to a target catalog alias in which the source description begins with the target alias description.

Source Description

Target Alias

Exema Persistent

Exema

Lexical to Alias TriGram

A string matching algorithm that compares tokens of 3-character length in source term descriptions to a target alias.

Lexical Trigram

A string matching algorithm that compares tokens of 3-character length in source and target term parsed lexical descriptions.

Phonetic

A word match where a source and target term descriptions match phonetically.

Source Description

Target Description

Total Proteen

Total Protein

Reverse Substring

A lexical match in which the target description begins with the source description.

Source Description

Target Description

Hyaluronate

Hyaluronate preparation

Reverse Included Substring

A lexical match in which the target description contains the source description.

Source Description

Target Description

Pregnancy Test

Unconfirm Pregnancy Test

SIFT Array

The SIFT array algorithm sends each term to a user-defined targeting-enabled SIFT array. The SIFT array’s targeting result terms are used as mapping candidates, with the same scoring and candidate schema used in the SIFT Array Test Harness.

Please Note: Unlike other algorithms, the SIFT Array algorithm only works when the map is checked out and, therefore, cannot be tested using the Update Candidates feature in Map Manager.

Stemmed Included Substring

A lexical match in which the stemmed source description contains the stemmed target description.

Source Description

Target Description

Tissue Tranglutaminas Antibody

Antibodies

Stemmed Reverse Included Substring

A lexical match in which the stemmed target description contains the stemmed source description.

Source Description

Target Description

Antibodies

Tissue Transglutaminas Antibody

Stemmed Reverse Substring

A lexical match in which the stemmed target description begins with the stemmed source description.

Source Description

Target Description

Immunoglobulins

Immunoglobulin Stimulated

Stemmed Substring

A lexical match in which the stemmed source description begins with the stemmed target description.

Source Description

Target Description

Immunoglobulin Stimulated

Immunoglobulins

Substring

A lexical match in which the source description begins with the target description.

Source Description

Target Description

diphtheria toxin preparations

diphtheria

Tokenized

A word match after each word in the term is sorted alphabetically.

Source Description

Target Description

aspirin/acetaminophen/caffeine

acetaminophen/aspirin/caffeine

Trigram

A string matching algorithm that compares tokens of 3-character length in source and target term descriptions. The spaces before and after the term description are taken into account and help broaden search results.

Referential

Community Consult Lexical

Utilizes the domain-specific percentage (Domain %) from a community map index package to return candidates and mapped targets based on threshold values where source terms have the same lexical values. Domain % refers to the summarized calculated data indicating the number of times a source term is mapped to a target code across all client nodes for a specific domain.

Algorithm Requirements: Your environment must be configured for community mapping. Additionally, you must be subscribed to the Community Map index package provided by your Community Map distributor through the Subscription Portal.

Community Consult Lexical (unrestricted)

Utilizes the unrestricted-domain percentage (All %) from a community map index package to return candidates and mapped targets based on threshold values where source terms have the same lexical values. All % refers to the summarized calculated data indicating the number of times a source term is mapped to a target code across all client nodes regardless of domain.

Algorithm Requirements: Your environment must be configured for community mapping. Additionally, you must be subscribed to the Community Map index package provided by your Community Map distributor through the Subscription Portal application.

Consult Map All External Target

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from a specified map where the source terms have the same external code values.

Consult Map All Domain Characteristics

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the specified map where the source terms have the same domain characteristic values.

Consult Map All Domain Lexical

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the specified map where the source terms have the same domain lexical values.

Consult Map All Domain Tokenized

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the specified map where the source terms have the same domain tokenized values.

Consult Map All Lexical

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the specified map where the source terms have the same lexical values.

Consult Map All Tokenized

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the specified map where the source terms have the same tokenized values.

Consult Map Domain Characteristics

Utilizes user-mapped terms (User or Symedical + User) from the specified map where the source terms have the same domain characteristic values.

Consult Map Domain Lexical

Utilizes user-mapped terms (User or Symedical + User) from the specified map where the source terms have the same domain lexical values.

Consult Map Domain Tokenized

Utilizes user-mapped terms (User or Symedical + User) from the specified map where the source terms have the same domain tokenized values.

Consult Map External Target

Utilizes the following to find external code (source) to source code (target) matches to source terms in another specified map with the same target catalog:

  • User-mapped terms (User or Symedical+User)

  • External code equals source code

The algorithm searches a single map. Therefore, when defining a map in Symedical Map Manager Algorithms application, selecting a single external target map consult algorithm searches only the assigned target map. Selecting one or more external target map algorithms searches all maps and controls the precedence in which each assigned target map is searched.

Consult Map Lexical

Utilizes user-mapped terms (User or Symedical + User) from the specified map where the source terms have the same lexical values. This algorithm searches a single map. Therefore, when defining a map in Symedical Map Manager Algorithms application, selecting a single consult map algorithm searches only the assigned target map. Selecting one or more consult map algorithms controls the precedence in which each assigned target map is searched.

Consult Map Tokenized

Utilizes user-mapped terms (User or Symedical + User) from the specified map where the source terms have the same tokenized values.

Consult Repository All Domain Characteristics

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the current map or any map with the same target catalog where the source terms have the same domain characteristic values.

Consult Repository All Domain Lexical

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the current map or any map with the same target catalog where the source terms have the same domain lexical values.

Consult Repository All Domain Tokenized

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the current map or any map with the same target catalog where the source terms have the same domain tokenized values.

Consult Repository All Identical Match

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the current map or any map with the same source and target catalog where the source terms match.

Consult Repository All Lexical

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the current map or any map with the same target catalog where the source terms have the same lexical values.

Consult Repository All Tokenized

Utilizes mapped terms (User, Symedical + User, or Symedical maps) from the current map or any map with the same target catalog where the source terms have the same tokenized values.

Consult Repository Domain Characteristics

Utilizes user-mapped terms (User or Symedical + User) from the current map or any map with the same target catalog where the source terms have the same domain characteristic values.

Consult Repository Domain Lexical

Utilizes user-mapped terms (User or Symedical + User) from the current map or any map with the same target catalog where the source terms have the same domain lexical values.

Consult Repository Domain Tokenized

Utilizes user-mapped terms (User or Symedical + User) from the current map or any map with the same target catalog where the source terms have the same domain tokenized values.

Consult Repository Identical Match

Utilizes user-mapped terms (User or Symedical + User) from the current map or any map with the same target catalog where the source terms match.

Consult Repository Tokenized

Utilizes user-mapped terms (User or Symedical + User) from the current map or any map with the same target catalog where the source terms have the same tokenized values.

Consult Repository Lexical

Utilizes user-mapped terms (User or Symedical + User) from the current map or any map with the same target catalog where the source terms have the same lexical values.

External Code to External Code

Code match on the external code of one term to the external code of another term within the source and target catalogs. If the same code type is included in the source and target catalogs, it can be used to assist in mapping.

External Code to Source Code

Code match on the external code of one term to the source code of another term. If an external code is included in the source catalog, that code is a source code in the target catalog.

Reciprocal Map

Symedical mapping is best match, one-way mapping. If a map exists in the opposite direction of the map being created, you may want to take advantage of the manual review and intervention that has been done on this opposite map in the analysis of your new map.

If so, you can specify that the Reciprocal Map algorithm be used for map analysis. This algorithm should be the low priority algorithm for your map, so that it is only used when a lexical type match is not found. Be sure to identify the reciprocal map in Symedical Map Manager Algorithms application, so that it is used when the map analysis is run.

For example, Map A maps Catalog 1 to Catalog 2 where:

Catalog 1 contains the following four terms:

  • (1) GINSENG ORAL TABLET

  • (2) GINSENG 100 mg ORAL CAPSULE (HARD, SOFT, ETC.)

  • (3) GINSENG ROOT 100 mg ORAL CAPSULE (HARD, SOFT, ETC.)

  • (4) COLD FIGHTER ORAL TABLET

and Catalog 2 contains the following two terms:

  • (5) Ginseng 100mg Oral capsule, liquid filled

  • (6) Pseudoephedrine Hydrochloride 120mg Oral tablet, extended release 12 hour

Map A maps Catalog 1 terms (1, 2, and 3) to Catalog 2 term (5). Additionally, a mapping of Catalog 1 term (4) to Catalog 2 term (6) is manually created. Now, create a map named Map B, which defines source Catalog 2 and target catalog 1.

The desired result is to have map analysis return a map from Catalog 2 term (5) to Catalog 1 term (2) and Catalog 2 term (6) to Catalog 1 term (4). The other two reciprocal maps for Ginseng do not map as well and shouldn't be considered until after the lexical and medication algorithm matches. The last mapping is only returned through the Reciprocal Map algorithm and should be reviewed as it was user indicated for the opposite direction.

Reciprocal Map by Incidence Count

Similar to the Reciprocal Map algorithm, the Reciprocal Map by Incidence Count algorithm reviews mappings from a specified map with the opposite source and target catalogs as the map the algorithm is being added to. However, candidates are displayed in descending order by their incidence count.

Reference Catalog Full Pivot

Returns candidates for a target based on how the source and target catalogs have been previously mapped to other catalogs in the repository. So, given a source catalog (a) and target catalog (b), if the source catalog (a) has been previously mapped to another catalog (c) and the target (b) has also been previously mapped to that catalog (c), then this algorithm can use that pre-existing map to go from (a) to (c) to (b), hence, pivoting off of the pre-existing relationships to catalog (c). This algorithm does that pivot for all catalogs that the source and target have in common.

Reference Catalog Pivot

Identical to the Reference Catalog Full Pivot algorithm with the exception that one specific map is specified as the map to consult.

Source Code to External Code

Code match on the source code of one term to the external code of another term. If a source code is included in the source catalog, that code is an external code in the target catalog.

Source Code to Source Code

Returns candidates based on an exact match between source and target catalog source codes.

Thesaurus Consult

Maps terms based on matching pivot codes (unique UMLS Metathesaurus or RxNorm codes). Each term's MT Code value is used to access the pivot code.

Algorithm Requirements:

  • A thesaurus must be updated within the UMLS Update application. Currently the RxNorm and UMLS Metathesaurus terminologies are supported and allow matching by comparing the MT Code values found in the source catalog with the NLM source terminology mnemonics (SABs) and NLM term types (TTYs) found in the chosen thesaurus. SABs and TTYs are defined for a thesaurus in the Terminology Source Manager application.

  • The source terms have MT Code values assigned.

Reference Map

External to External Reference Map

Uses the selected reference map’s source and target external code types to match source and target terms. The algorithm searches a single reference map. Therefore, when defining a map in Symedical Map Manager Algorithms application, selecting one map algorithm searches only the selected map. Selecting this algorithm controls the precedence in which each assigned target map is searched. If the algorithm is configured to return candidate mappings, candidates appear under Additional Symedical Candidates in the Map Manager Candidates tab.

External to Source Reference Map

Searches the specified reference map by external code for candidates. It searches the specified reference map for source term value matches to source term external code value, and returns candidates whose linked target code value matches the target catalog source code value. If the algorithm is configured to return candidate mappings, candidates appear under Additional Symedical Candidates in the Map Manager Candidates tab.

Source to External Reference Map

Searches the specified reference map by source code for candidates. It searches the specified reference map for matching source code and returns candidates whose linked target code value matches the target catalog external code value. If the algorithm is configured to return candidate mappings, candidates appear under Additional Symedical Candidates in the Map Manager Candidates tab.

Source to Source Reference Map

Searches the specified reference map by source code for candidates. It searches the specified reference map for matching source term values and returns candidates whose linked target code value matches the target catalog source code value. If the algorithm is configured to return candidate mappings, candidates appear under Additional Symedical Candidates in the Map Manager Candidates tab.