=Paper=
{{Paper
|id=Vol-2518/paper-ODLS2
|storemode=property
|title=LABO: An Ontology for Laboratory Test Prescription and Reporting
|pdfUrl=https://ceur-ws.org/Vol-2518/paper-ODLS2.pdf
|volume=Vol-2518
|authors=Adrien Barton,Paul Fabry,Luc Lavoie,Jean-François Ethier
|dblpUrl=https://dblp.org/rec/conf/jowo/BartonFLE19
}}
==LABO: An Ontology for Laboratory Test Prescription and Reporting==
https://ceur-ws.org/Vol-2518/paper-ODLS2.pdf
LABO: An Ontology for Laboratory Test
Prescription and Reporting
Adrien BARTONa,b,1, Paul FABRYb, Luc LAVOIEb and Jean-François ETHIERb,1
a
Institut de Recherche en Informatique de Toulouse (IRIT), CNRS, France
b
Groupe de Recherche Interdisciplinaire en Informatique de la Santé (GRIIS),
Université de Sherbrooke, Quebec, Canada
Abstract. LABO is an ontology formalizing laboratory test prescriptions and
reporting documents. It is built according to the OBO Foundry methodology, and is
a component of a core ontological model that aims to enable interoperability
between various clinical data sources in the context of a Learning Health System.
This article presents LABO, distinguishing between directive entities and data items,
and using the relations directs and is about to represent their connections with the
relevant laboratory test processes.
Keywords. Laboratory test, Information content entity, Directive information entity
1. Introduction
Learning Health Systems analyze health information generated from patients in order to
provide secondary use of clinical data and decision support. They rely on access to a
wide range of clinical data, such as drug prescriptions or laboratory test prescriptions and
results, usually scattered across numerous heterogeneous information systems.
Applied ontologies can support a common, source-independent representation of
these data, thus helping to solve the “Tower of Babel problem” in medical informatics.
An ontology has already been developed for drug prescriptions: the Prescription of
DRugs Ontology (PDRO, read “Pedro”) [1]. This paper presents the creation of an
ontology using a compatible methodology for representing laboratory test prescriptions
and reporting documents: LABO (for LABoratory Ontology). It is being used in the
context of the Canadian PASS architecture (“Plateforme apprenante en soutien aux
systèmes de santé et services sociaux”), which is used in several projects including
PARS3 (“Plateforme apprenante pour la recherche en santé et services sociaux au
Québec”), an ontology-based Learning Health System that builds on the former proof of
concept European project TRANSFoRM [2]. Ontologies such as PDRO and LABO are
being used to generate a relational database structure [3]. This structure is then mapped
to databases from various healthcare institutions, in order to support a system of data
mediation on the model of what was done in TRANSFoRM [2].
1
Corresponding Authors. IRIT, avenue de l’étudiant, 31400 Toulouse, France. GRIIS, Université de
Sherbrooke, 2500, boul. de l’Université, Sherbrooke (Québec), J1K 2R, Canada. E-mails:
adrien.barton@irit.fr; jf.ethier@usherbrooke.ca.
Copyright © 2019 for this paper by its authors. Use permitted under Creative Commons License
Attribution 4.0 International (CC BY 4.0).
Some terminologies such as portions of LOINC [4] or SNOMED-CT [5] have been
developed to represent laboratory tests. LABO does not aim at represent the variety of
possible laboratory tests (which is rather the domain of the OBO Foundry’s Ontology for
Biomedical Investigations OBI [6]), but instead at representing the structure of the
documents that are generated to prescribe them and report on them. However, codes from
LOINC or SNOMED-CT are commonly used in various health databases, and could
therefore be used in the general relational database built on the basis of LABO (with a
system mapping synonymous codes from various terminologies, as decribed in [7]).
2. Methods
LABO has been developed according to a realist approach based on the Basic Formal
Ontology (BFO), as a candidate to the OBO Foundry [8]. Like other OBO Foundry
candidates, it re-uses other OBO Foundry classes to keep the ontologies compatible with
each other. In the following, classes names will be prefixed by the ontology name when
they are imported from another OBO Foundry candidate ontology. LABO introduces 87
new classes (some of which via sub-ontologies we developed of demographic or
administrative informational entities), and imports whole or parts of OBO Foundry
ontologies such as the Information Artifact Ontology (IAO) [9], the Ontology for
Biomedical Investigations (OBI) [6] and the Ontology for General Medical Science
(OGMS) [10]. The ontology can be found at the following address:
https://github.com/OpenLHS/LABO.
LABO represents informational entities that direct laboratory tests or report on them,
and their parts, as subclasses of IAO:Information content entity (“ICE”) [9]. In particular,
it represents entities directing laboratory tests as subclasses of IAO:Directive information
entity (“DIE”) [6] and laboratory results as instances of IAO:Data item. It is based on
BFO and provides Aristotelian definitions for the created classes [11].
Following the OBO Foundry methodology, we also reuse object properties from
other OBO Foundry ontologies and avoid introducing new ones when possible. Still, we
introduced a new object property directs (and its inverse directed by) that has as domain
IAO:Directive information entity and as range BFO:Process. The formalization of this
relation is ongoing in other work [12]: in the present article, we will provide an informal
characterization to ensure proper comprehension of the core of this work, by stating that
d directs p means that an agent represents the DIE d in his cognitive system, has the
intention to follow it, and follows it as a consequence of this intention.
3. Results
3.1. A Use Case Scenario
Let us consider the following scenario S1. Dr. Jones wants to know more about the health
status of his patient Mr. Fiennes. He requests several lab tests for him: a complete blood
count (CBC), as well as a serum sodium measurement. On a request form (which can be
named a “prescription” by analogy with drug prescriptions), he writes ‘CBC, serum
sodium’ (that will be called ‘LADIG0’). He adds the name of Mr. Fiennes, the date of
the day (‘May 31st, 2019’), and signs it. Several tests are then realized on Mr. Fiennes as
a result of this prescription: a serum sodium test, and a dozen of distinct tests that are
directed by the instruction ‘CBC’: a hematocrit test, a hemoglobin concentration test, etc.
On June 5th, 2019, he receives a paper stating ‘serum sodium: in progress; CBC in
progress’. On June 8th, 2019, he receives another paper stating ‘serum sodium: 138
mmol/L; CBC in progress’. On June 15th, 2019, he receives a final report giving the value
of all the tests that were prescribed by LADIG0. LABO provides categories for all those
and related entities, as pictured on figures 1 and 2 below and explained in the following.
IAO: Information content entity
IAO:Document
PDRO:Health care prescription
Laboratory test prescription
Laboratory test report document
IAO:Directive information entity
IAO:Action specification
Laboratory test directive item ("LADI")
Laboratory test directive group ("LAD group")
PDRO:Condition [see taxonomy on figure 2]
Laboratory test reporting information
Laboratory test reporting item ("LAR item")
Laboratory test result item
Laboratory test reporting group ("LAR group")
Laboratory test reporting item time specification
Laboratory test status specification
IAO:Data item
OGMS:Clinical finding
Laboratory result2
Specimen characteristic specification
BFO:Process
OGMS:Health care process assay
Laboratory test3
Directed laboratory test group
Figure 1. Taxonomy of relevant entities and abbreviations
2
The class Laboratory finding is currently defined by OGMS as “A representation of a quality of a
specimen that is the output of a laboratory test and that can support an inference to an assertion about some
quality of the patient.” However, a successful laboratory test can lead to an ICE that is about an entity that is
not a quality of the specimen, such as e.g. a disposition or a coagulation time. Therefore, we introduced the
class ‘Laboratory result’, which we define as “A representation of an entity related to a specimen that is the
output of a laboratory test.”.
3
The class Laboratory test is currently defined by OGMS as “A measurement assay that has as input a
patient-derived specimen and as output a data item that is about a quality of the specimen.” However, we think
that it should rather be defined as “A measurement assay that has as input a specimen derived from an organism
and that aims at having as output a data item that is about an entity related to the specimen.” As a matter of
fact, additionally to the problem mentioned in the former footnote, a laboratory test can fail in having as output
such a data item if e.g. the sample is spoiled.
3.2. Laboratory Test Direction
3.2.1. Laboratory Test Directive Item
A central class in LABO is Laboratory test directive item (abbreviated hereafter “LADI”)
which is a subclass of IAO:Action specification, and is defined as “An action
specification that directs one or several laboratory tests and such that none of its proper
parts directs some but not all of those laboratory tests.” This definition is motivated by
the fact that some instructions, such as ‘CBC ’, direct several distinct tests – more than a
dozen for the complete blood count: hematocrit, hemoglobin, etc.; but no part of the ICE
‘CBC ’ does specifically direct a hematocrit test, or a hemoglobin test, etc. (although a
proper part of ‘CBC ’, namely ‘CBC’ – note the absent space at the end – directs the
same laboratory tests). Therefore, both ‘CBC’ and ‘serum sodium’ mentioned above are
instances of LADI. On the other hand, the mereological sum ‘CBC; serum sodium’ is not
a LADI, as it has two parts (‘CBC’ and ‘serum sodium’) that each direct some test(s).
Several LADIs can be gathered into a Laboratory test directive group (“LAD
group”), defined as “An action specification that has as members one or several
laboratory test directive items, as well as possibly some statements specifying a starting
condition, a stopping condition and a testing condition.” In scenario S1 above, ‘CBC,
serum sodium’ is a LAD group.
A LAD group is always composed of at least one LADI:
LAD group subClassOf Action specification and has part some LADI
A LAD group might be composed of only one LADI: if Dr. Jones would not have
prescribed a CBC to Mr. Fiennes, but only the serum sodium test, then ‘serum sodium’
would be both a LADI and a LAD group.
Finally, we define a Laboratory test prescription as “A health care prescription
specifying the realization of one or several laboratory test(s). A laboratory test
prescription encompasses at least one laboratory test directive item group.”:
Laboratory test prescription subClassOf PDRO:Health care prescription
and has part some LAD group
A Laboratory test prescription might have several LAD groups though; consider for
example a prescription with the instructions:
‘CBC on 2019/06/01;
Na, K, creatinine on 2019/08/01’
Here, ‘CBC’ is a first LAD group, and ‘Na, K, creatinine’ is a second LAD group.
Laboratory test prescription inherits the following axiom from PDRO:Heatlh care
prescription:
Laboratory test prescription subClassOf
has part some IAO:Author identification
and has part some PDRO:Patient identification
and has part some PDRO:Document creation time identification
An example of laboratory test prescription is the prescription mentioned above that has
LADIG0 as part. Many LADIs are parts of a Laboratory test prescription, but not all of
them – consider e.g. an instruction directing a laboratory test as part of a research study.
3.2.2. Laboratory Test Conditions
As suggested by the definition of LAD group provided above, some LAD groups can
have a conditional structure. Consider for example ‘PTT q2h start 2h post-op for 24h’,
where ‘PTT’ stands for “partial thromboplastin time”, ‘q2h’ for “every 2 hours”, and
‘post-op’ stands for “post operation”. It is composed by the LADI ‘PTT’, as well as a
Starting laboratory test protocol condition ‘2h post-op’, a Continuing laboratory test
protocol condition ‘for 24h’ (which is synonymous to an ending condition being satisfied
24h after starting) and a Laboratory test administration condition ‘q2h’. As it happens,
this is a similar structure to a drug prescription structure such as ‘Amoxicilin 500 mg q8h
for 14 days, start in case of symptoms of bronchitis’. And indeed, very similar classes
are already defined in PDRO, such as Starting drug administration condition, Continuing
drug administration condition and Dosing condition. We therefore introduce the parent
classes: Starting condition, Continuing condition and Action condition, with the
following taxonomy:
PDRO:Condition
Starting condition
Starting laboratory test protocol condition
PDRO:Starting drug administration condition
Continuing condition
Continuing laboratory test protocol condition
PDRO:Continuing drug administration condition
Action condition
Laboratory test administration condition
PDRO:Dosing condition
Figure 2. Extract from the taxonomy of conditions in PDRO and LABO
3.3. Laboratory Tests
Although LABO focuses on informational entities pertaining to lab test prescription and
reporting, the connection between those informational entities and the laboratory tests
they are related to needs to be represented. As a matter of fact, it can be important to
represent the connection between several ICEs – e.g., representing that several ICEs were
generated by the same laboratory test. A LADI does not necessarily direct a Laboratory
test (think about a LADI that is never followed), but any Laboratory test is directed by
some LADI (at least the LADI that is concretized in the brain of the person performing
the lab test, if we follow IAO’s ontology of mental entities [9]); therefore, we add the
following axiom:
Laboratory test subClassOf OGMS:Health care process assay
and directed by some LADI
We also need to introduce the entity Directed laboratory test group as “A health
care process assay constituted by all laboratory test(s) that are directed by a single
laboratory test directive item.” (where Health care process assay is defined in OGMS
[10]). In our example, the item ‘CBC’ above directs an instance of Directed laboratory
test group composed by an instance of Hematocrit test, an instance of Hemoglobin test,
etc. Thus, all the laboratory tests that are directed by the same LADI are parts of one
Directed laboratory test group (which is also directed by this LADI):
Directed laboratory test group subClassOf OGMS:Health care process assay
and directed by exactly 1 LADI
and has part some Laboratory test
3.4. Laboratory Test Reporting
3.4.1. Laboratory Test Reporting Item
Once a laboratory test has been performed, one or several documents may report on its
progress and result. Consider again scenario S1. Suppose that on 2019/06/15, Dr. Jones
receives a document with the date, the name of his patient, and results about the tests that
were prescribed, such as:
• ‘RBC 5.2 10^12/L, 2019/06/15’ (where ‘RBC’ stands for “Red blood cell
count”)
• ‘WBC 12.1 10^9/L, 2019/06/15’ (‘WBC’ stands for “White blood cell count”)
• ‘serum sodium 138 mmol/L, 2019/06/08’
• etc.
All those ICEs and some of their parts – such as the result ‘WBC 12.1 10^9/L’, the
date ‘2019/06/08’, or the status report ‘in progress’, are instances of Laboratory test
reporting information (“LAR information”), which is defined as an ‘An information
content entity which reports on some aspect(s) of a particular laboratory test.’
More specifically, we define a Laboratory test reporting item (“LAR item”) as “A
laboratory test reporting information that is part of a laboratory test reporting group, that
is about a laboratory test, and that mentions which characteristic of a specimen this test
was supposed to assess, and a time at which this information was valid”. For example,
‘WBC 12.1 10^9/L, 2019/06/15’ and ‘RBC 5.2 10^12/L, 2019/06/15’ are LAR items.
Note that a LAR item does not necessarily report a result of a test: for example,
‘serum sodium in progress, 2019/06/05’ is a LAR item. To account for such items, we
define a Laboratory test status specification as “A laboratory test reporting information
that specifies the status of a group of laboratory tests.” (where “a group of laboratory
tests” might refer to only one laboratory test) Examples of laboratory test status
specifications include ‘in progress’ (before a test result was obtained), ‘resulted’ (when
a test result was obtained by a machine or a technician, but was not validated yet by the
person in charge), ‘validated’ (when the test result was validated by the person in charge)
or ‘canceled’.
A reporting information must contain a time specification and an information
specifying the specimen characteristic to be of enough relevance. Therefore, we define a
Laboratory test reporting item time specification (“LAR item time specification”) as “A
laboratory test reporting information that specifies a time at which a laboratory test
reporting item was valid.” This time specification does not refer to the time when a given
lab document was generated, but rather the moment at which information from the
laboratory test process was created as it unfolded. Examples of LAR item time
specifications would be ‘2019/06/08’ or ‘2018/07/12’ mentioned above.
Finally, we define a Specimen characteristic specification as “An information
content entity that specifies a particular characteristic of a specimen or a class of
characteristics of specimens.” For example, ‘blood group’ in a prescription of blood
group test or ‘serum sodium concentration’ in a LAR item ‘serum sodium concentration
140 mmol/L, validated, 2018/07/28’ are instances of Specimen characteristic
specification.
Once those classes are defined, we propose the following necessary conditions for a
LAR item: it is about a Laboratory test, and it contains ICEs specifying the time and
specimen characteristic (an additional axiom is mentioned below in 3.4.3):
Laboratory test reporting item subClassOf
Laboratory test reporting information
and has part exactly 1 LAR item time specification
and has part exactly 1 Specimen characteristic specification
and is about some Laboratory test
3.4.2. Laboratory Test Result Item
As we mentioned above, not all LAR items include a laboratory result. To characterize
the LAR items that do include a result, we define Laboratory test result item as “A
laboratory test reporting item that includes the result of the laboratory test” and
Laboratory result as “A clinical finding representing an entity related to a specimen that
is the output of a laboratory test”:
Laboratory test result item subClassOf Laboratory test reporting item
and has part some Laboratory result
Laboratory result subClassOf OGMS:Clinical finding
and is specified output of some OGMS:Laboratory test
For example, in scenario S1 above, ‘serum sodium 138 mmol/L’ is an instance of
Laboratory test result item, and ‘138 mmol/L’ is an instance of Laboratory result4.
3.4.3. Laboratory Test Reporting Group and Report Document
As we defined it earlier, a LAR item is about only one test. Therefore, a LAR information
such as ‘CBC in progress’ is not a LAR item, because it reports on several tests, not just
one. To provide a more specific class for such instructions than LAR information, we
define Laboratory test reporting group (“LAR group”) as “A laboratory test reporting
information which reports on the test(s) directed by one laboratory test directive item.”
– that is, it reports on the Directed laboratory test group directed by this LADI. For
example, reporting information such as ‘serum sodium: 138 mmol/L’, ‘CBC in progress’
or the whole list of results for a unique CBC (‘CBC: RBC 5.2 10^12/L, 2019/06/15;
WBC 12.1 10^9/L, 2019/06/15 [etc.]’) are each instances of LAR group.
On the other hand, a LAR item is always part of a LAR group (even if this group is
only composed by this LAR item):
Laboratory test reporting item subClassOf
Laboratory test reporting information
and part of some Laboratory test reporting group
4
Note that this does not mean that all ICEs that are composed by the chain of characters “138 mmol/L”
are instances of Laboratory result: such informational entities can be created outside the context of a laboratory
test. However, we endorse here the conception presented in [9], according to which the identity of an ICE
depends on the intention of the creator of this ICE. Therefore, the ICE ‘138 mmol/L’ created in scenario S1 is
indeed an instance of Laboratory result.
Since a LAR group reports on the Directed laboratory test group, it is about it. Also, a
LAR group has as part the LADI that directed the Directed laboratory test group that this
LAR group is about (for example, ‘CBC in progress’ contains the LADI ‘CBC’):
Laboratory test reporting group subClassOf
Laboratory test reporting information
and is about some Directed laboratory test group
and has part some Laboratory test directive item
Finally, we define a Laboratory test report document as ‘A document reporting on
one or several laboratory tests’. It has as part (at least) one LAR Group:
Laboratory test report document subClassOf IAO:Document
and has part some Laboratory test reporting group
Note that not all laboratory tests lead to a laboratory result (think about a test that does
not conclude to any result because, for example, the sample was spoiled).
Figure 3. Example of a laboratory test report document concerning a CBC 5
4. Discussion and Conclusion
The LABO ontology formalizes laboratory test prescriptions, results and reporting, as
well as their parts. Along with PDRO, it is a part of a core ontological model to enable
interoperability between various clinical data sources in a LHS context: data at different
levels of taxonomical generality or mereological extent can be annotated using the
various classes of those two ontologies.
Future work will investigate more in detail the structure of laboratory test results,
which can be given in a variety of formats (scalar, ratio, intervals with inclusive or
exclusive boundaries, etc.). An important point (already noticed in [1], p. 285, fn 6) is
that several classes introduced in LABO seem to imply a role character. For example,
5
In addition to the laboratory test report items mentioned above, we usually find information such as the
range of normal values for a given test and a flag indicating if a test is abnormal. These items have also been
represented in the ontology.
some ICE particulars instantiate Laboratory result or Starting condition because they
bear some roles: ‘140 mmol/L’ mentioned above is a laboratory result because it has
been generated by a process of a specific kind – that is, because it stands in some relation
with other entities (in particular, processual entities); similarly, ‘at t, symptoms of anemia
are present’ would be a Starting condition because it is part (say) of a laboratory test
protocol. Therefore, more detailed theoretical investigations in how ICEs can bear roles
and the identity of ICEs will be required.
Acknowledgments
Many thanks are due to François Goyer, Olivier Grenier and Christina Khnaisser for their
precious inputs in the creation of this ontology, as well as to three anonymous reviewers
for some very detailed and helpful comments. A poster presenting the basis of LABO
has been presented at ICBO 2019 (10th International Conference in Biomedical
Ontology) [13].
References
[1] J.-F. Ethier, A. Barton, R. Taseen, An ontological analysis of drug prescriptions, Applied Ontology 13
(2018), 273–294.
[2] J.-F. Ethier, M. McGilchrist, A. Barton, A.-M. Cloutier, V. Curcin, B.C. Delaney, et al., The
TRANSFoRm project: Experience and lessons learned regarding functional and interoperability
requirements to support primary care, Learning Health Systems (2017),.
[3] C. Khnaisser, L. Lavoie, B. Fraikin, A. Barton, A. Burgun, J.-F. Ethier, Generating a relational database
for heterogeneous data using an ontology., Draft.
[4] C.J. McDonald, S.M. Huff, J.G. Suico, G. Hill, D. Leavelle, R. Aller, et al., LOINC, a universal standard
for identifying laboratory observations: a 5-year update, Clinical Chemistry 49 (2003), 624–633.
[5] K. Donnelly, SNOMED-CT: The advanced terminology and coding system for eHealth, Studies in
Health Technology and Informatics 121 (2006), 279.
[6] A. Bandrowski, R. Brinkman, M. Brochhausen, M.H. Brush, B. Bug, M.C. Chibucos, et al., The ontology
for biomedical investigations, PloS One 11 (2016), e0154556.
[7] J.-F. Ethier, O. Dameron, V. Curcin, M.M. McGilchrist, R.A. Verheij, T.N. Arvanitis, et al., A unified
structural/terminological interoperability framework based on LexEVS: application to TRANSFoRm,
Journal of the American Medical Informatics Association 20 (2013), 986–994.
[8] B. Smith, M. Ashburner, C. Rosse, J. Bard, W. Bug, W. Ceusters, et al., The OBO Foundry: coordinated
evolution of ontologies to support biomedical data integration, Nature Biotechnology 25 (2007), 1251–
1255.
[9] B. Smith, W. Ceusters, Aboutness: Towards foundations for the information artifact ontology, in:
Proceedings of the 6th International Conference on Biomedical Ontology, CEUR Workshop Proceedings,
Lisbon, Portugal, 2015, pp. 1–5.
[10] R.H. Scheuermann, W. Ceusters, B. Smith, Toward an ontological treatment of disease and diagnosis,
in: Proceedings of the 2009 AMIA Summit on Translational Bioinformatics, San Francisco CA, 2009, pp.
116–120.
[11] R. Arp, B. Smith, A.D. Spear, Building Ontologies with Basic Formal Ontology, The MIT Press, 2015.
[12] A. Barton, L. Vieu, J.-F. Ethier, Directing actions, in: Proceedings of the Joint Ontology Workshops
2019, CEUR Workshop Proceedings, Graz, Austria, accepted, pp. 1–10.
[13] P. Fabry, A. Barton, F. Goyer, O. Grenier, J.-F. Ethier, First steps towards an ontology for laboratory test
prescription and reporting, in: Proceedings of the 10th International Conference on Biomedical Ontology
(ICBO 2019), Buffalo, NY, US, accepted, p. 1.