=Paper=
{{Paper
|id=Vol-2050/odls-paper4
|storemode=property
|title=An Ontology for Clinical Laboratory Standard Operating Procedures
|pdfUrl=https://ceur-ws.org/Vol-2050/ODLS_paper_4.pdf
|volume=Vol-2050
|authors=Fatima S. Maikore,Gantigmaa Selenge,Adebola Olayinka,Pamela Abbott,Larisa Soldatova
|dblpUrl=https://dblp.org/rec/conf/jowo/MaikoreSOAS17
}}
==An Ontology for Clinical Laboratory Standard Operating Procedures==
https://ceur-ws.org/Vol-2050/ODLS_paper_4.pdf
An Ontology for Clinical Laboratory
Standard Operating Procedures
Fatima S. MAIKOREa,b,1, Gantigmaa SELENGEa, Adebola OLAYINKAc, Pamela
ABBOTT d, and Larisa SOLDATOVAa
a
Brunel University, London, UK
b
Baze University, Abuja, Nigeria
c
Ahmadu Bello University, Zaria, Nigeria
d
The University of Sheffield, Sheffield, UK
Abstract. This paper reports on a knowledge representation model, OCL-SOP, that
formally defines key entities pertinent to clinical laboratory procedures. OCL-SOP
provides a formal description of laboratory experimental actions, biochemical
entities involved, equipment used, input and output data, and the data processing
actions. We developed OCL-SOP in collaboration with domain experts. We
demonstrate the utility of this knowledge model through a mobile application,
SmartSOP, which gives laboratory technicians access to digitized laboratory
standard operating procedures (SOPs) and enables searching through the
semantically annotated SOPs. OCL-SOP is available at:
github.com/fatibaba/EXACT-med/blob/master/OCL-SOP.owl
Keywords: knowledge representation, ontology, clinical laboratory procedures
1.Introduction
Laboratory testing in hospitals is a crucial step in the diagnosis of diseases. It is, therefore
important to ensure quality and reduce error rates in the laboratory testing process.
Exchange of laboratory test results, which forms part of the testing process, is another
important factor to consider in laboratories. The discrepancies in the measurement
standards, terminologies, reporting formats, and interpretation of test results between
different laboratories complicate the exchange and integration of clinical information [1].
Health organisations have come together to develop Standard Operating Procedures
(SOP) in an attempt to standardise laboratory practices in hospitals. Laboratories use
these SOPs for "correct test selection, sample collection and handling, while standardized
test terminology, and units of traceability to ISO standard 17511 are required to ensure
equivalency of measurement results” [2]. However, the availability of guidance
documents does not guarantee their use in the laboratories [3]. One of the major problems
faced by practitioners is the inconveniences of finding required information quickly in
the SOP documents. These documents are provided as free text either in PDF or MS
Word formats with pages ranging from 12 to more than 50. The laboratory technicians
find navigating through such documents time consuming, and the search interferes with
1
Corresponding Author, Computer Science Department, Baze University, PMB264, Plot 686 COO,
Kuchigoro, Abuja, Nigeria; E-mail: fatima.maikore@brunel.ac.uk, fatima.baba@bazeuniversity.edu.ng.
the actual testing process. Another problem is the representation of SOPs in free text
using non-standardized terminology which leads to difficulties in the computational
comparison of procedures and also in the reproducibility of the results [4].
We propose a knowledge representation model, OCL-SOP (an Ontology for Clinical
Laboratory SOP), that formally defines laboratory procedures and the associated data.
Ontologies have been useful for knowledge representation in different areas including
healthcare [5-7] and are becoming popular for representing clinical practice guidelines
[8]. We extensively analysed Standards for Microbiology Investigations (SMI), that are
SOPs provided by the Nation Institute for Health and Care Excellence (NICE) [9], and
also consulted with microbiology experts for the development of OCL-SOP.
OCL-SOP aims to facilitate clinical data exchange, support information retrieval and
improve reproducibility of laboratory procedures. We demonstrate the utility of this
knowledge model through a mobile application, SmartSOP. SmartSOP gives laboratory
technicians access to digitised SOPs, enabling semantic information retrieval, recording
test results in a standardized form and recording relevant activities in the lab.
2.Related Works
Several projects are focusing on the semantic representation of biomedical protocols.
BioAssay Ontology (BAO) describes chemical biology screening assays and their results
for the purpose of categorising assays [10]. Another example of standardized protocol
representation is the Ontology for Biomedical Investigations (OBI) [11] which represents
different phases and activities involved in biomedical investigations. “OBI broadly
describes terms that are applicable to the biomedical and technological domain which
address the need for a cross-disciple ontology” [11].
The ontology for experimental actions (EXACT) provides a generic semantic
representation of experimental protocols to ensure their reproducibility by humans and
machines [12]. EXACT describes such experimental actions as 'add', 'incubate' along
with their descriptors, e.g. 'temperature', 'biochemical entities' involved, and 'period'.
The described projects do not focus on protocols that are specific to hospital
laboratories. In this paper, we present OCL-SOP which re-uses relevant representations,
and principally extends EXACT by knowledge pertinent to clinical laboratory protocols.
3.OCL-SOP
We developed OCL-SOP to capture the descriptions of the laboratory experimental
actions, biochemical entities involved, equipment used, input and output data, and also
the data processing actions. We developed OCL-SOP in collaboration with domain
experts; they were involved in the specification of requirements and validation of the
descriptions of experimental actions. We focused on NICE SMIs, with a total of 47
currently available. We extracted from these SMIs all the terms used in the description
of experimental actions and their descriptors.
All experimental actions defined in EXACT were present in the considered SMIs.
Therefore, we used EXACT as the upper-level ontology for OCL-SOP and imported
other suitable classes from other sources. The structure of OCL-SOP inherits the EXACT
ontology structure, and has the following main branches:
• 'Process' to model processual entities, e.g. 'experimental action'.
• 'information content entity' to model information about SOPs, e.g. 'date
submitted'.
• 'Descriptor of experimental action' to model entities pertinent to the actions, e.g.
an action 'incubate' requires the description of a 'period' of incubation, and
what 'biochemical entity' will be incubated.
• 'Optional descriptor of experimental action' to model non-essential (but still
useful) entities pertinent to actions, e.g. 'temperature' may not be essential for
the description of 'incubate' since a default value of 30 may be used.
Figure 1 shows a fragment of the OCL-SOP classes and their hierarchy. The class
names are prefixed with their source ontology name to show which ones are imported
and which ones are OCL-SOP specific classes.
Figure 1. The upper-level of the OCL-SOP.
We refined the descriptors of several experimental actions. Our analysis shows that
the NICE SMIs express ‘temperature’ and ‘period’ as ranges. To capture this, we created
such descriptors as 'min temperature', 'max temperature', 'min period', and 'max period'.
The analysis of NICE SMIs shows that there are many data-specific actions, see
Figure 2 with a fragment of the urine investigation SMI document with data-specific
actions 'estimate' and 'record'. To capture data-specific actions, we extended OCL-SOP
by adding the 'data action' sub-branch to the 'process' branch. In EXACT, the actions
'record', 'measure', 'calculate', and 'count' were defined as 'experimental actions'. We
re-modelled these actions as data-specific actions. We also added new 'data actions', e.g.
‘convert’ and ‘estimate’.
Figure 2. A fragment of urine analysis SMI showing data actions.
We identified 68 new experimental actions, i.e. actions that were not defined in any
other knowledge representation resource. They were added to OCL-SOP as new classes,
e.g. ‘emulsify’ and ‘double dilute’. The semantic meaning of the classes in OCL-SOP is
specific to the clinical laboratory domain. For example, ‘double dilute’ means to do a
serial dilution of a solution, and this differs from the conventional meaning in other areas:
horses carrying a certain type of gene. Table 1 shows examples of new terms in OCL-
SOP and their definitions.
Table 1. Examples of new terms and their definitions in OCL-SOP.
Term Name OCL-SOP Branch Definition
‘double dilute’ Experimental Action A dilution of a solution to reduce its concentration made twice.
estimate Data Action To give a range of probable values for a given entity.
tilt Experimental Action To slant a biochemical entity or equipment slightly to one side.
stain Experimental Action To use a colouring agent to enhance the structure of a
‘biochemical entity’, usually a specimen, for easy viewing,
using equipment such as a microscope
The terminology in SMIs differs from the terminology used in biological protocols.
To capture that, we added SMI-specific synonyms to terms previously defined in other
resources. For example, the synonym ‘agitate’ was added to the term ‘shake’ via the
relation ‘has-synonym’.
Many of these terms have been already defined in other knowledge models. For
example, 'incubate' and 'mix' are defined in EXACT, and 'decant' is defined in the
National Cancer Institute Thesaurus (NCIT) [13]. Following the best practices in
ontology development and the recommendations by OBO foundry [14], we re-used these
definitions and imported the relevant classes to OCL-SOP. We used OntoMaton [15], an
ontology search tool, to find ontologies in the BioPortal that have defined similar terms.
Table 2 shows example re-used classes and the source ontologies.
Table 2. Examples of re-used terms in OCL-SOP.
Term Name Ontology Source Ontology Full Name
count EXACT An Ontology for Experimental Actions
decant NCIT National Cancer Institute Thesaurus
observe BIOMO Biological Observation Matrix Ontology
streak EXACT An Ontology for Experimental Actions
dispense NCIT National Cancer Institute Thesaurus
We developed OCL-SOP using Protégé, and it is freely available in OWL-DL
format with all external import files at https://github.com/fatibaba/EXACT-med.
4.Use Case: SmartSOP
We used the terms defined in OCL-SOP ontology to represent Urine Investigation SMI
and the Internal Quality Assurance (IQA) procedure of a particular microbiology
laboratory from a hospital in London. This laboratory needs a simple representation of
SOPs to encourage technicians to follow the guidance and allow them to retrieve the
necessary procedural information quickly. We developed a mobile app SmartSOP that
enables an easy access to the lab procedures. SmartSOP provides a user friendly
semantically annotated versions of the Urine Investigation SMI and IQA.
The use of OCL-SOP ensures that all the terms are disambiguated (via globally
unique identifiers) and have logically consistent definitions. The semantic search
functionality enables users to find information about experimental actions easily,
including by their synonyms. SmartSOP enables lab technicians to record the results of
lab procedures in a dedicated MySQL database, and export these results to CSV files.
This makes sharing of the results by different labs easier despite the differences in the
platforms used. SmartSOP provides lab technicians with a secure login and keeps track
of the activities of each technician. This is important for quality assurance management
in the laboratory.
Clinical lab practitioners and biomedical scientists evaluated our SmartSOP
positively. We conducted a user acceptance test; six laboratory technicians used the app
and their feedback was captured through a questionnaire. The technicians agreed that it
would make their work easier and more efficient. They suggested that such an app will
be even more useful for the representation of more complex procedures. Such procedures
require specialist knowledge in order for them to be performed due to the vast number
of rules and conditions lab technicians must consider, and the complexity of interpreting
results. A usability evaluation of the app in accordance with the usability principles
provided by the Health Information Management Systems Society that measure
effectiveness, efficiency and user satisfaction of medical applications, was also positive.
5.Conclusion and Future Work
In this paper, we report on the development of OCL-SOP, an ontology for the
representation of hospital laboratory SOPs. OCL-SOP imports relevant terms from
several knowledge models and defines 68 new terms required for the representation of
NICE SMIs. We present a mobile application SmartSOP which utilises OCL-SOP for
the visualisation of SOPs, semantic search, and recording clinical test results.
We plan to analyse more clinical laboratory SOPs to provide a better coverage of
actions in OCL-SOP. We are working on aligning OCL-SOP with a newly emerging
standard, the Robot Task Ontology [16] to support the development of autonomous
robots which will carry out clinical laboratory procedures. We will add more complex
laboratory procedures such as the detection of Carbapenemases to the SmartSOP
application. This procedure includes complex algorithms which are more difficult to
follow. We also plan to support automatic update of SOPs in SmartSOP whenever OCL-
SOP is updated.
Acknowledgements
The work reported in this paper was partially funded by the CHIST-ERA AdaLab
project (EPSRC UK grant EP/M015661/1).
References
[1] M. Plebani and M. Panteghini, Promoting clinical and laboratory interaction by harmonization, Clinica
Chimica Acta 432 (2014), 15-21.
[2] J.R. Tate et al., Harmonization of laboratory testing current achievements and future strategies, Clinica
Chimica Acta 432 (2014), 4-7.
[3] A.L. Francke et al., Factors influencing the implementation of clinical guidelines for health care
professionals: a systematic meta-review, BMC medical informatics and decision making 8(1) (2008), 1.
[4] R.R. Brinkman et al., Modeling biomedical experimental processes with OBI, J. Biomedical Semantics
1(S-1) (2010), 7.
[5] D. Riao et al., An ontology-based personalization of health-care knowledge to support clinical decisions
for chronically ill patients, Journal of biomedical informatics 45(3) (2012), 429-446
[6] J. Kim and K.-Y. Chung, Ontology-based healthcare context information model to implement ubiquitous
environment, Multimedia Tools and Applications 71(2) (2014), 873-888.
[7] Balubaid, M.A. and U. Manzoor, Ontology Based SMS Controller for Smart Phones. arXiv preprint arXiv
1509.01379 (2015).
[8] A. Galopin et al., An Ontology-Based Clinical Decision Support System for the Management of Patients
with Multiple Chronic Disorders, In MEDINFO 2015: EHealth-enabled Health: Proceedings of the 15th
World Congress on Health and Biomedical Informatics (2015).
[9] www.gov.uk/government/collections/standards-for-microbiology-investigations-smi
[10] A. M. Clark et al., BioAssay templates for the semantic web. PeerJ Computer Science 2 (2016), e61.
[11] R. R. Brinkman et al., Modeling Biomedical Experimental Processes with OBI, Journal of Biomedical
Semantics 1(Suppl 1) (2010), S7.
[12] L. N. Soldatova et al., The EXACT description of biomedical protocols, Bioinformatics 24(13) (2008)
i295-i303.
[13] http://bioportal.bioontology.org/ontologies/NCIT?p=summary
[14] http://www.obofoundry.org/principles/fp-000-summary.html
[15] https://github.com/ISA-tools/OntoMaton
[16] S. Balakirsky et al., Towards A Robot Task Ontology Standard, In Proceedings of the ASME 2017
International Manufacturing Science and Eng. Conference (2017).