Health Information modelling language - overview: Difference between revisions

From Endeavour Knowledge Base
Line 41: Line 41:
== What the language must do==
== What the language must do==
Health data can be conceptualised as a graph, and thus the model of health data is a graph model.   
Health data can be conceptualised as a graph, and thus the model of health data is a graph model.   
A graph is considered as a set of nodes and edges. For a graph to be valid, a node must have at least one edge, and an edge must be connected to at least two nodes. Thus the smallest practical graph must have at least 3 entities.


The language must be both human and machine readable in text form.
The language must be both human and machine readable in text form.
Line 48: Line 46:
The language must use the recognisable plain language characters in UTF-8. For human readability the characters read from left to right and for machine readability a graph is a character stream from beginning to end.
The language must use the recognisable plain language characters in UTF-8. For human readability the characters read from left to right and for machine readability a graph is a character stream from beginning to end.


It is hard to produce a machine readable language that humans can understand.


Consequently two grammars are developed, one for human legibility and the other for optimised machine processing. However, both must be human and machine readable and translators are a prerequisite.
Two grammars are required; one for human legibility, and the other for optimised machine processing. However, both must be human and machine readable.


A model presented in the human legible grammar must be translatable directly to the machine representation without loss of semantics. In the ensuing paragraphs the human optimised grammar is illustrated but in the final language specification both are presented side by side to illustrate semantic translatability.
A model presented in the human legible grammar must be translatable directly to the machine representation without loss of semantics. In the ensuing paragraphs the human optimised grammar is illustrated but in the final language specification both are presented side by side to illustrate semantic translatability.


'''Semantic triples'''
'''Example'''
 
Consider a health record about a person holding the date of birth.
 
http://xyz/dateOfBirth might be considered a property that holds a value that is a date of birth. Subjects and objects may also have identifiers which may or may not be meaningful. As subjects and objects operate as nodes, and nodes require edges,  predicates thus assume the role of edges in a graph.
 
To make sense of the language, subjects require constructs that include predicate object lists, object lists, and objects which can themselves be subjects defined by predicates. Put together with the terse language requirement, the grammar used in the human oriented language is simplified form of TURTLE. TURTLE itself is directly compatible with all of the Discovery grammars. The following snip illustrates the main TURTLE structures together with the punctuation:
<pre style="background-color:#fcfaee">
Subject1
  Predicate1 Object1;
  Predicate2 Object2;                # predicate object list separated by ';'
  Predicate3 Object3,
              Object4,
              Object5;              #object list
  Predicate4 [                        #anonymous object with predicates enclosed []
                Predicate5 Object6;
                Predicate6 Object7
              ]
.                                      # end of sentence
 
Subject2 Predicate1 Object1.            #Simple triple                   
</pre>
 
 
'''Semantic context'''
 
The interpretation of a structure is often dependent on the preceding predicate. Because the language is semantically constrained to the profiles of the sublanguages, certain punctuations can be semantically interpreted in context. For example, as the semantic axioms incorporates OWL EL but not OWL DL, Object Intersection (and) is supported but not Object union. In other words, in certain contexts there are ANDS but not ORS.


This allows for the use of a collection construct, for example in the following equivalent definitions of a grandfather, in the first example the grandfather is an equivalent to an intersection of a person and someone who is male and has children, and the second one is an intersection of a person, something that is male, and someone that has children. Both interpretations assume AND as the operator because OR is not supported at this point in  OWL EL.
Consider them definition of a grandfather, in the first example the grandfather is an equivalent to a person who is male and has children who are people that have children.
<syntaxhighlight lang="turtle" style="border:3px solid grey">
<syntaxhighlight lang="turtle" style="border:3px solid grey">
Grandfather
Grandfather

Revision as of 15:06, 13 February 2021

This article describes a language used for modelling data used in health records as well as the means by which health record queries can be defined in a system independent manner.

For those familiar with the semantic web languages, it is safe to assume that the language described herein is simply a dialect (or profile) of the standard W3C recommendations of RDF, RDFS, OWL2, SHACL and SPARQL. The profile is designed to simplify implementations of the model and to constrain it’s features to ensure relevance and optimum performance.

The language is designed to support an information model conceived as a graph model and thus implementation of the model and related health data are represented logically as a graph.


Purpose background and rationale

Question: Yet another language? Surely not.

Answer: No, not at all. What can be stated though is that it is not a health domain language. It is a language used commonly across all sectors across the world, applied to the health and social care domain.

One of the ambitions of the language is to free the representation of health data from the modelling silos within healthcare itself. This is based on the principle that the only difference between health data and other data is the vocabulary and ontologies used.

The following sections first describe the purpose of a modelling language, the background to the Endeavour /Discovery approach and the rationale behind the approach adopted.

Subsequently the sections break down the various aspects of the language at ever increasing granularity, emphasising the relationship between the language fragments and the languages from which they are derived, resulting in the definition of the grammar of the language. The relationship between the language and a model store is also described.

Purpose of the language.

The main purpose of a modelling language is to exchange data and information about information models in a way that both machines and humans can understand. A language must be able to describe the types of structures that make up an information model as can be seen on the right. Diagrams and pictures are all very well, but they cannot be used by machines.

It is necessary to support both human and machine readability so that a model can be validated both by humans and computers. Humans can read diagrams or text. Machines can read data bits. The two forms can be brought together as a stream of characters forming a language.

A purely human based language would be ambiguous, as all human languages are. A language that is both can be used to promote a shared understanding of often complex structures whilst enabling machines to process data in a consistent way.

It is almost always the case that a very precise machine readable language is hard for humans to follow and that a human understandable language is hard to compute consistently. As a compromise, many languages are presented in a variety of grammars and syntaxes, each targeted at different readers. The language in this article all adopt a multi-grammar approach in line with this dual purpose.

Multi grammars are the norm in computer languages. Different software implementations use different technologies and different grammars are needed. The crucial point is that the different grammars mean the same thing in the end, just presented in different ways with different syntaxes.

Contributory languages

An information model can be modelled as a Graph i.e. a set of nodes and edges (nodes and relationships, nodes and properties). Likewise, health data can be modelled as a graph conforming to the information model graph.

The world standard approach to a language that models graphs is RDF, which considers a graph to be a series of interconnected triples, a triple consisting of the language grammar of subject, predicate and object. Thus the modelling language uses RDF as its fundamental basis, and can therefore be presented in the RDF grammars. The common grammars used in this article include TURTLE (terse RDF Triple language) and JSON-LD (json linked data) which enables simple JSON identifiers to be contextualised in a way that one set of terms can map directly to internationally defined terms or IRIs.

RDF in itself holds no semantics whatsoever i.e. it is not practical to infer or validate or query based purely on an RDF structure. To use RDF it is necessary to provide semantic definitions for certain predicates and adopt certain conventions. In providing those semantic definitions, the predicates themselves can then be used to semantically define many other things.

The three aspects alluded to above are covered by the logical inclusion of W3C semantic based languages, described further in the sublanguages section of this article.


What the language must do

Health data can be conceptualised as a graph, and thus the model of health data is a graph model.

The language must be both human and machine readable in text form.

The language must use the recognisable plain language characters in UTF-8. For human readability the characters read from left to right and for machine readability a graph is a character stream from beginning to end.


Two grammars are required; one for human legibility, and the other for optimised machine processing. However, both must be human and machine readable.

A model presented in the human legible grammar must be translatable directly to the machine representation without loss of semantics. In the ensuing paragraphs the human optimised grammar is illustrated but in the final language specification both are presented side by side to illustrate semantic translatability.

Example

Consider them definition of a grandfather, in the first example the grandfather is an equivalent to a person who is male and has children who are people that have children.

Grandfather
   isEquivalentTo 
              Person,                             #class 1
              [ hasGender Male;                   #class 2
                hasChild Person,                 #class 2.1
                          [hasChild Person]  ]     #class 2.2  
.

Which can also be illustrated as:

or Grandfather
  isEquivalentTo Person,                            #class 1
                   hasGender Male,                   #class  2
                   hasChild Person,                 #class 3
                             [hasChild Person]     #class 4
 ].

Machine oriented grammar

JSON is a popular syntax currently and thus this is used as an alternative.

JSON represents subjects , predicates and objects as object names and values with values being either literals or or objects.

JSON itself has no inherent mechanism of differentiating between different types of entities and therefor JSON-LD is used. In JSON-LD identifiers resolve initially to @id and the use of @context enables prefixed IRIs and aliases.

The above Grandfather can be represented in JSON-LD (context not shown) as follows:

{"@id" : "Grandfather",
 "EquivalentTo" :[{ "@id":"Person"},
                  {"hasGender": {"@id":"Male"}},
                  {"hasChild": [{"@id":"Person"},
                                {"hasChild" : {"@id":"Person"}}]]}}

Which is equivalent to the verbose version 2 syntax showing explicit intersections and classes defined only by object property

{"iri" : "Grandfather",
 "EquivalentTo" :[
    {"Intersection":[
       { "Class": {"iri":"Person"}},
       {"ObjectPropertyValue": {
           "Property": {"iri":"hasGender"},
            "ValueType": {"iri":"Male"}
            }},
      {"ObjectPropertyValue": {
           "Property":{"iri":"hasChild"},
            "Expression":{
                "Intersection":[ 
                   {"Class":{"iri":"Person"}},
                   {"ObjectPropertyValue":{
                      "Property":{"iri": "hasChild"},
                      "ValueType": {"iri":"Person"}} ] } } ]}


Language sublanguage grammar and syntax

Sub languages

The Discovery language, as a mixed language, has its own grammars , but in addition a number of W3C recommended languages can be used in their respective grammars and syntaxes for the elements of interest. This enables multiple levels of interoperability. For example, the information model contains an OWL2 ontology and can therefore be accessed via any of the OWL2 syntaxes. The information model also contains data models in the form of shapes and can therefore be accessed via SHACL. Queries can be exchanged via SPARQL and expression constraints by ECL.

For those who want a consistent syntax encompassing the entire model then Discovery grammar can be used and this is supported via a simplified TURTle, TURTLE itself and JSON-LD as well as JSON serialised java classes that reflect some standard object oriented objects.

Many specialised sublanguages overlap with others in a way that makes them difficult to translate to each other. For example, in ECL, the reserved word MINUS (used to exclude certain subclasses from a superclass) , could be mapped to the much more obscure OWL2 syntax that requires the modelling of class IRIs "punned" as individual IRIs, in order to properly exclude instances when generating lists of concepts. Likewise it could be mapped to SPARQL. The Discovery information model helps resolve this problem by supporting the sub languages directly.

The following sublanguages are supported in TURTLE, JSON-LD or OWL Functional syntax:

  • OWL2, which is used for semantic definition and inference. In line with convention, only OWL2 EL is used and thus existential quantification and object intersection can be assumed in its treating of class expressions and axioms. The open world assumption inherent in OWL means it is very powerful for subsumption testing but cannot be used for constraints without abuse of the grammar.
  • SHACL, which is used for data modelling constraint definitions. SHACL can also include OWL constructs but its main emphasis is on cardinality and value constraints. It is an ideal approach for defining logical schemas, and because SHACL uses IRIs and shares conventions with other W3C recommended languages it can be integrated with the other two aspects. Furthermore, as some validation rules require quite advanced processing SHACL can also include query fragments.
  • SPARQL, GRAPHQL are both used for query. SPARQL forms the basis of interoperable query. GRAPH QL, when presented in JSON-LD is a pragmatic approach to extracting graph results via APIs as its type and directive systems enables properties to operating as functions or methods. SPARQL is a more standard W3C query language for graphs but suffers from its own in built flexibility (and an ambiguous issue with subqueries) making it hard to produce consistent results. Consequently a pragmatic SPARQL profile is supported. The degree of SPARQL is included to the extent that it can be easily interpreted into SQL or other query languages. SPARQL with entailment regimes are in effect SPARQL query with OWL support.
  • RDF itself. RDF triples can be used to hold objects themselves and an information model will hold many objects which are instances of the classes as defined above (e.g. value sets and other instances)

The information modelling services used by Discovery can interoperate using the above sub-languages, but Discovery also includes a language superset making it easy to integrate. For example it is easy to mix OWL axioms with data model shape constraints as well as value sets without forcing a misinterpretation of axioms.

Foundation grammars and syntaxes

Discovery language has its own Grammars built on the foundations of the W3C RDF grammars:

  • A terse abbreviated language, TURTLE
  • SPARQL for query
  • JSON-LD representation, which can used by systems that prefer JSON, wish to use standard approaches, and are able to resolve identifiers via the JSON-LD context structure.
  • Proprietary JSON based object serializable grammar. This directly maps to the internal class structures used in Discovery and can be used by client applications that have a strong contract with a server.

Identifiers aliasing and context

Concepts are identified and referenced by the use of International resource identifiers (IRIs).

Identifiers are universal and presented in one of the following forms:

  1. Full IRI (International resource identifier) which is the fully resolved identifier encompassed by <>
  2. Abbreviated IRI a Prefix followed by a ":" followed by the local name which is resolved to a full IRI
  3. Aliases. The core language tokens (that are themselves concepts) have aliases for ease of use. For example rdfs:subClassOf is aliased to subClassOf,

There is of course nothing to stop applications using their own aliases and when used with JSON-LD @context may be used to enable the use of aliases.

Data is considered to be linked across the world, which means that IRIs are the main identifiers. However, IRIs can be unwieldy to use and some of the languages such as GRAPH-QL do not use them. Furthermore, when used in JSON, (the main exchange syntax via APIs) they can cause significant bloat. Also, identifiers such as codes or terms have often been created for local use in local single systems and in isolation are ambiguous.

To create linked data from local identifiers or vocabulary, the concept of Context is applied. The main form of context in use are:

  1. PREFIX declaration for IRIs, which enable the use of abbreviated IRIs. This approach is used in OWL, RDF turtle, SHACL and Discovery itself.
  2. VOCABULAR CONTEXT declaration for both IRIs and other tokens. This approach is used in JSON-LD which converts local JSON properties and objects into linked data identifiers via the @context keyword. This enables applications that know their context to use simple identifiers such as aliases.
  3. MAPPING CONTEXT definitions for system level vocabularies. This provides sufficient context to uniquely identify a local code or term by including details such as the health care provider, the system and the table within a system. In essence a specialised class with the various property values making up the context.

The following is an example of the use of the prefix directives for IRIs and to define aliases for some of the owl and rdfs tokens

@prefix :  <http://www.DiscoveryDataService.org/InformationModel/Ontology#>.
@prefix rdf:  <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix sh:   <http://www.w3.org/ns/shacl#> .
@prefix xsd:  <http://www.w3.org/2001/XMLSchema#> .
@prefix owl:  <http://www.w3.org/2002/07/owl#>.
@prefix sn: <ttp://snomed.info/sct#>.


rdfs:subClassOf :alias ("subClassOf").

owl:Class  :alias ("Class").

:alias
   a owl:DatatypeProperty;
   <http://www.DiscoveryDataService.org/InformationModel/Ontology#alias> 
   ("alias").

rdfs:label
    a owl:annotationProperty;
    :alias ("name" "label") .

owl:DataTypeProperty
    a owl:DataTypeProperty;
    :alias ("dataProperty") .

As the aliases are defined they can now be used in an abbreviated Turtle syntax, as well as the standard turtle syntax. For example, the following Snomed-CT code is defined as a class and a subclass of another Snomed-CT code.

sn:407708003 a Class; name : "Sample appearance (observable entity)"; subClassOf sn:407708003.

A JSON-LD equivalent of the above uses context to cover prefixes and aliases in a simpler manner

"@context" : {
   
    "alias" : {
      "@id" : "http://www.DiscoveryDataService.org/InformationModel/Ontology#alias",
      "@type" : "@id"
    },
    "@base" : "http://www.DiscoveryDataService.org/InformationModel/Ontology#",
    "" : "http://www.DiscoveryDataService.org/InformationModel/Ontology#",
    "rdf" : "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
    "sh" : "http://www.w3.org/ns/shacl#",
    "owl" : "http://www.w3.org/2002/07/owl#",
    "xsd" : "http://www.w3.org/2001/XMLSchema#",
    "rdfs" : "http://www.w3.org/2000/01/rdf-schema#"
  }

Resulting in the standard JSON-ld context based approach:

 {"@id" : "sn:12345",
    "@type" : "owl:Class",
    "subClassOf" : "sn:34568"
  }

Ontology structures and vocabulary

For the purposes of reasoning the semantic ontology axiom and class expression vocabulary uses the tokens and structure from the OWL2 profile OWL EL, which itself is a sublanguage of the OWL2 language

However, in addition some standard OWL2 DL axioms are used in order to provide a means of specifying additional relationships that are of value when defining relationships. The following table lists the main owl types used and example for each. Note that their aliases are used for brevity. Please refer to the OWL2 specification to describe their meanings

Owl construct usage examples
Class An entity that is a class concept e.g. A snomed-ct concept or a general concept
ObjectProperty 'hasSubject' (an observation has a subject that is a patient)
DataProperty 'dateOfBirth' (a patient record has a date of birth attribute
annotationProperty 'description' (a concept has a description)
SubClassOf Patient is a subclass of a Person
Equivalent To Adverse reaction to Atenolol is equivalent to An adverse reaction to a drug AND has causative agent of Atenolol (substance)
Disjoint with Father is disjoint with Mother
Sub property of has responsible practitioner is a subproperty of has responsible agent
Property chain is sibling of'/ 'is parent of' / 'has parent' is a sub property chain of 'is first cousin of'
Inverse property is subject of is inverse of has subject
Transitive property is child of is transitive
Existential quantification Chest pain and

Finding site of - {some} thoracic structure

Object Intersection Chest pain is equivalent to pain of truncal structure AND finding in region of thorax AND finding site of thoracic structure
Individual All chest pain subclasses but not the specific instance of acute chest pain
DataType definition Date time is a restriction on a string with a regex that allows approximate dates
Property domain a property domain of has causative agent is allergic reaction
Property range A property range of has causative agent is a substance


Use of Annotation properties for original codes

Annotation properties are the properties that provide information beyond that needed for reasoning.  They form no part in the ontological reasoning, but without them, the information model would be impossible for most people to understand. Annotation properties can also be used for implementation supporting properties such as release status, version control, authoring dates and times and so on. 

Typical annotation properties are names and descriptions. They are also used as meta data such as a status of a concept or the version of a document.

Many concepts are derived directly from source systems that used them as codes, or even free text.

The concept indicates the source and original code or text (or combination) in the form actually entered into the source system. It should be noted that many systems do not record codes exactly as determined by an official classification or provide codes via mappings from an internal id.  It is the codes or text used from the publishers perspective that  is used as the source.

Thus in many cases, it is convenient to auto generate a code, which is then placed as the value of the “code” property in the concept, together with the scheme. From this, the provenance of the code can be inferred.

Each code must have a scheme. A scheme may be an official scheme or  proprietary scheme or a local scheme related to a particular sub system.

For example, here are some scheme/ code combinations

Scheme Original Code/Text/Context Concept code/ Auto code Meaning
Snomed-CT            47032000         47032000         Primary hydrocephaly
EMIS- Read H33-1   H33-1   Bronchial asthma
EMIS – EMIS EMISNQCO303 EMLOC_EMISNQCO303 Confirmed corona virus infection
Barts/Cerner Event/Order=687309281 BC_687309281 Tested for Coronavirus (misuse of code term in context)
Barts/Cerner Event/Order= 687309281/ResultTxt= SARS-CoV-2 RNA DETECTED BC_dsdsdsdx7 Positive coronavirus result

 

Note that in the last example, the original code is actually text and has been contextualised as being from the Cerner event table, the order field having a value of 687309281 and the result text having a value of ResultTxt= SARS-CoV-2 RNA DETECTED

Shape structures and vocabulary

As in the semantic ontology, the language borrows the constructs from the W3C standard SHACL, which can also be represented in any of the RDF supporting languages such as TURTLE or JSON-LD.


Data mapping

This part of the language is used to define mappings between the data model and an actual schema to enable query and filers to automatically cope with the ever extending ontology and data properties. 

This is part of the semantic ontology but uses the idea of context (described later on).

Query

It is fair to say that data modelling and semantic ontology is useless without the means of query.

The current approach to the specification of query uses the GRAPHQL approach with type extensions and directive extensions.

Graph QL , (despite its name) is not in itself a query language but a way of representing the graph like structure of a underlying model that has been built using OWL. GRAPH QL has a very simple class property representation, is ideal for REST APIs and results are JSON objects in line with the approach taken by the above Discovery syntax.

Nevertheless, GRAPHQL considers properties to be functions (high order logic) and therefore properties can accept parameters. For example, a patient's average systolic blood pressure reading could be considered a property with a single parameter being a list of the last 3 blood pressure readings. Parameters are types and types can be created and extended.

In addition GRAPHQL supports the idea of extensions of directives which further extend the grammar.

Thus GRAPHQL capability is extended by enabling property parameters as types to support such things as filtering, sorting and limiting in the same way as an.y other query language by modelling types passed as parameters. Subqueries are then supported in the same way.

GRAPHQL itself is used when the enquirer is familiar with the local logical schema i.e. understands the available types and fields. In order to support semantic web concepts an extension to GRAPHQL, GRAPHQL-LD is used, which is essentially GRAPH-QL with JSON-LD context.

GRAPH QL-LD has been chosen over SPARQL for reasons of simplicity and many now consider GRAPHQL to be a de-facto standard. However, this is an ongoing consideration.

ABAC language

Main article : ABAC Language

The Discovery attribute based access control language is presented as a pragmatic JSON based profile of the XACML language, modified to use the information model query language (SPARQL) to define policy rules. ABAC attributes are defined in the semantic ontology in the same way as all other classes and properties.

The language is used to support some of the data access authorisation processes as described in the specification - Identity, authentication and authorisation .

This article specifies the scope of the language , the grammar and the syntax, together with examples. Whilst presented as a JSON syntax, in line with other components of the information modelling language, the syntax can also be accessed via the ABAC xml schema which includes the baseline Information model XSD schema on the Endeavour GitHub, and example content viewed in the information manager data files folder