=Paper=
{{Paper
|id=Vol-2477/paper_6
|storemode=property
|title=G-PROV: Provenance Management for Clinical Practice Guidelines
|pdfUrl=https://ceur-ws.org/Vol-2477/paper_6.pdf
|volume=Vol-2477
|authors=Nkechinyere N. Agu,Neha Keshan,Shruthi Chari,Oshani Seneviratne,James P. McCuske,Deborah L. McGuinness
|dblpUrl=https://dblp.org/rec/conf/semweb/AguKCSMM19
}}
==G-PROV: Provenance Management for Clinical Practice Guidelines==
https://ceur-ws.org/Vol-2477/paper_6.pdf
Copyright ©2019 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
G-PROV: Provenance Management for Clinical
Practice Guidelines
Nkechinyere N. Agu1[0000−0003−1386−8602] , Neha Keshan1[0000−0002−0752−6406] ,
Shruthi Chari1[0000−0003−2946−7870] , Oshani Seneviratne1[0000−0001−8518−917X] ,
Sabbir M. Rashid1[0000−0002−4162−8334] , Amar K. Das2[0000−0003−3556−0844] ,
James P. McCusker1[0000−0003−1085−6059] , and Deborah L.
McGuinness1[0000−0001−7037−4567]
1
Rensselaer Polytechnic Institute, Troy, NY 12180, USA
2
IBM Research, Cambridge, MA
Abstract. Providing provenance of treatment suggestions made by clin-
ical decision support systems can enhance transparency and trust in
these systems by healthcare practitioners. Provenance can aid in resolv-
ing ambiguity and conflicts between various guideline sources. We have
developed a guideline provenance ontology, G-Prov, by extending exist-
ing provenance ontologies, to enable accurate encoding of the source of
the reasoning rules that decision support systems rely on to generate
diagnosis and treatment suggestions. Our ontology enables provenance
representation at different granularity levels within guidelines. For in-
stance, G-Prov can be used to annotate rules with citations found in
evidence sentences as well as other sources of knowledge, such as figures
and tables. Additionally, we have developed an application to show a
range of use cases for our ontology. We demonstrate our work annotat-
ing recommendations in a CPG for Type-2 Diabetes and discuss how
our approach could be used in various medical settings where CPGs are
utilized.
Keywords: Provenance · Clinical Practice Guidelines · Clinical Deci-
sion Support Systems · Ontology · Biomedical Knowledge Graphs
Ontology: https://purl.org/heals/gprov
Website: https://tetherless-world.github.io/GProv/
1 Introduction
Guideline-based decision support systems aim to assist healthcare practitioners
with patient diagnosis and treatment choices. These systems embody informa-
tion present in authoritative clinical practice guidelines (CPGs) as computable
knowledge (e.g. logical rules). However, most of these systems fail to provide the
source of treatment suggestions. Furthermore, the information used to generate
these treatment suggestions might become obsolete, or worse yet, invalidated,
within a newer guideline. Recording the provenance of treatment suggestions
68
2 N. Agu et al.
helps address the challenging task of systematically updating a system when new
guidelines or medical literature are published. To concretely tackle the problem
of lack of provenance, there needs to be an easy way to extract information
within the CPG and associate them as provenance.
We attach provenance to SWRL [5] treatment rules of the Diabetes Mellitus
Treatment Ontology (DMTO) [4]. However since DMTO’s rules lack provenance,
we use DMTO as an example usage of our guideline provenance ontology: G-
Prov and annotate these rules with recommendations from the 2018 version
of the ADA Standards of Medical Care guideline (ADA CPG)3 . DMTO is an
ontology that provides treatment suggestions for Type-2 Diabetes. They use in-
formation from several medical guidelines on Diabetes, including the ADA CPG,
Diabetes Canada,4 and the European Association for the study of Diabetes.5 As
stated earlier, DMTO lacks information on the source of each rule within the
ontology, making it difficult to evaluate the accuracy of and the evidence for
each rule. Hence, we address this issue, in our paper.
To this end, we develop a guideline provenance ontology, G-Prov, to encode the
provenance details in CPGs. For our ontology, we reused several existing ontolo-
gies, such as, W3C provenance ontology (PROV-O) [7], Dublin Core Metadata
Terms (DCT) [6], and the Bibliographic Ontology (BIBO) [2]. Further, utilzing
concepts from G-Prov, we modeled the provenance of DMTO rules with content
from the ADA CPGs as Resource Description Framework (RDF) [9] Knowledge
Graphs (KGs). We show the diverse use of our ontology-enabled approach by
developing an application for three different use cases that cover a variety of
users of guideline documents.
2 Related Work
Several general-purpose provenance ontologies currently exist such as the Prove-
nance Ontology (PROV-O) [7], which is an ontology that provides classes and
properties to capture generic provenance terms in various domain. The Dublin
Core Metadata Terms (DCT) [6], which is a lightweight vocabulary that aids
in the description of resources and provides a list of terms that can be used to
describe metadata information. DCT can be used to represent articles, figures,
tables on a higher level. On the other hand, The Bibliographic Ontology (BIBO)
[2], serves as an enhancement of the DCT ontology and contains a more detailed
description of referenced articles. However, all these provenance ontologies ad-
dress the issue of modeling of provenance at a more general level. In G-Prov,
we build off these foundational provenance ontologies to associate provenance of
clinical decisions with their authoritative guideline evidence.
The clinical domain have plethora of heterogenous data of varying qualities.
Hence, there have been several efforts to create ontologies that enable increased
traceability, transparency and trust in clinical data. ProvCaRe [12] is an ontology
3
ADA Standards of Medical Care: http://care.Diabetesjournals.org/
4
Diabetes Canada: https://guidelines.Diabetes.ca/cpg
5
EASD: https://www.easd.org/
69
G-PROV: Provenance Management for Clinical Practice Guidelines 3
designed to enhance reproducibility of scientific research by capturing the meta-
data of published articles. Their ontology focuses on the details of the scientific
study including the design description, the data collection methods, analysis of
the data and overall research methodology. Provenance Context Entity (PaCE)
[11] is a an ontology that enables provenance tracking in scientific data. This
ontology has been applied to the Biomedical Knowledge Repository to integrate
biomedical data from various biomedical literature, databases, and vocabulary
lists. Furthermore, in healthcare, there is an ontology for enabling provenance in
healthcare management [3] that enables the efficient integration of patient data
from various clinical sources, and helps in understanding patient outcome as a
result of a clinical process. While these ontologies exist for the clinical domain,
they fail to capture provenance terms necessary for guideline-specific data. The
authors of [13] have developed a semantic web system that attempts to capture
guideline interactions. Their research has some similarities with our work, how-
ever, they focus on identifying causation which is not within the scope of G-Prov
ontology.
3 Guideline Provenance Ontology
We aim to provide an ontology that is compatible with best practice, rele-
vant provenance ontologies, and which is focused at directly support clinical,
guideline-based decision support systems. Our initial attempts focus mainly on
annotating SWRL rules defined within DMTO ontology with the recommenda-
tions found within the ADA CPG. Typically, a CPG comprises of recommen-
dations for clinical practice, cited publications backing these recommendations
etc. To address the use case of associating provenance with treatment sugges-
tions encoded as rules, we included classes and properties (described shortly in
section 3.1) in the G-Prov ontology, to link these rules to different granularities
of evidence within CPGs.
3.1 Main Classes and Property Associations
Our ontology is represented in OWL-DL and was developed using Protege [10].
The G-Prov ontology 6 comprises of terms broadly belonging to three sections.
A broader provenance section that captures metadata associated with the prove-
nance of a treatment rule in the context of its backing CPG recommendation.
These terms were obtained from the general purpose ontologies described earlier
in section 2, i.e. PROV-O, DCT, and BIBO. The second section is comprised of
more domain specific terms that are used to associate rules to their guideline
evidence. The third section discusses the guideline specific classes and properties.
6
We use the following ontology prefixes: (1) gprov: G-Prov (2) sio: SemanticScience
Integrated Ontology (3) bibo: The Bibliographic Ontology (4) prov: PROV-O: The
PROV ontology (5) dct: Dublin Core Terms (6) sco: Study Cohort Ontology (7) do:
Disease Ontology (8) ncit: National Cancer Institute Thesaurus
70
4 N. Agu et al.
3.2 Provenance in Guidelines
Both CPGs (ncit:C28237) and the cited publications within them are specific
sub-classes of bibo:Document. Additionally, citations within CPGs are repre-
sented by gprov:Citation, and are linked to their reference information using
prov:hasPrimarySource. Further, every published document has at least one au-
thor and, to capture the list of all the authors in a publication, we use the
class bibo:Author, which contains one or more authors. We also capture other
metadata information associated with a document using terms from both BIBO
(bibo:volume, bibo:issue, bibo:pageStart, bibo:pageEnd, bibo:uri) and DCT (dct:creator,
dct:title) ontologies.
3.3 Associating provenance with treatment rules
Since we are annotating treatment rules, we need a class to model them and we
introduce the gprov:FormalRule class. Additionally, these rules (gprov:FormalRule)
were obtained from guideline recommendations (gprov:Recommendation) and we
use the prov:wasDerivedFrom property to link the two classes. At a more generic
level, we can also link the rules directly to the guideline (ncit:C28237). We use
dct:partOf association to represent the relationship between a recommendation
and its host CPG.
Fig. 1. A simple instance diagram showing the modeling of a recommendation within
the ADA CPG using G-Prov Ontology
3.4 Guideline Specific Classes and Properties
CPGs are developed to manage a specific health condition. In G-Prov, we link the
guideline class (ncit:C28237) to the health condition (gprov:DiseaseManagement)
71
G-PROV: Provenance Management for Clinical Practice Guidelines 5
it addresses via prov:used property. The gprov:DiseaseManagement class links
to (via prov:used) at least one specific disease type (do:Disease). Additionally,
some guidelines carry a measure of the quality of each recommendation, and we
capture this information using gprov:Grade. Furthermore, guideline recommen-
dations are backed by one or more evidence sentences (gprov:EvidenceSentence)
within the guideline. These evidence sentences contain citations, whose modeling
using the existing provenance ontologies is explained in section 3.2.
4 ADA Standards of Care Guideline Knowledge Graph
There are numerous CPGs for effective diagnosis and treatment of diseases.
Through G-Prov, we support the modeling of recommendations in the ADA
CPG. The ADA CPG is aimed at providing treatment recommendations for
healthcare practitioners to help in the management of Type-1 and Type-2 Dia-
betes. The ADA CPG is released annually, with updates to recommendations,
recommendation grade, evidence supporting each recommendation, and the med-
ical literature that backs each piece of evidence.
To annotate a treatment rule with its ADA CPG recommendation, we first did
a manual pass on the CPG to understand its structure. Subsequently, we wrote
a web scraping script using BeautifulSoup7 to extract all the contents of the
guideline into a JSON file. The contents of the JSON file included all the recom-
mendations, their grades, possible evidence sentences, and the citations within
each sentence. Our initial efforts focused on two chapters from the ADA CPG,
namely the “Pharmacologic Approaches to Glycemic Treatment” and “Cardio-
vascular Disease and Risk Management” chapters.
To construct the KG, we created a spreadsheet to organize the extracted guide-
line data. Further, we manually identified applicable guideline recommendations
for DMTO treatment rules and mapped them within the spreadsheet. We had
our medical domain expert (also a co-author) go over these annotations to con-
firm both completeness and accuracy. Finally, we wrote a SETLr [8] script to
automatically convert the contents of the spreadsheet to an RDF KG. Fig. 1
highlights the resulting KG from modeling a recommendation within the ADA
CPG using G-Prov ontology.
5 Applications of G-Prov
The G-Prov ontology and the ADA KG were used to develop an application and
showcase its various uses for a variety of users. The application evolved out of the
need for healthcare practitioners to identify the sources of treatment suggestions
made by CDS. The various pieces of information that a healthcare practitioner
needed to make clinical decisions were identified with the help of our medical
domain expert and we began iteratively building an ontology-enabled system for
7
https://www.crummy.com/software/BeautifulSoup
72
6 N. Agu et al.
our use case. We used a standard ontology engineering process to ensure both
completeness and adaptability into a wide range of applications. Our application
captures three main use cases, described in the next three paragraphs. Screen-
shot of the various views of the applications can be viewed on our resources
website 8 .
The first use case focuses on providing provenance information of treatment
Fig. 2. A high-level work flow architecture diagram for the guideline provenance ap-
plication
suggestions made by CDS, including the source of the information, other med-
ical literature that support the article and the date of publication. Thus, when
a treatment suggestion is made by a CDS system, the healthcare practitioner’s
view enables the healthcare practitioner to query the system for more informa-
tion about its suggestion.
The second use case focuses on a Population Health Manager (PHM). To create
a view for this user, we integrate the Study Cohort Ontology [1], to assist in
visualizing information about citations. Through this view, we aim to help the
PHM identify subgroups (cohorts) of patients who may benefit from the recom-
mendations. Additionally, the visualizations also help them analyze the research
publications that serve as evidence for recommendations.
The final use case is focused on a content developer. This view would allow for
developers of CDS systems to enter provenance information while creating/edit-
ing the decision rules. The added advantage of creating a content developers view
in our application, over other editing tools such as Protege, is that this view does
not assume the user has any semantic knowledge whatsoever. This view includes
a form to collect information from the user in a friendly, easy to read manner
and creates a KG of the information entered using the G-Prov ontology.
8
Visit https://tetherless-world.github.io/GProv/
73
G-PROV: Provenance Management for Clinical Practice Guidelines 7
6 Results and Discussion
We have developed a guideline provenance ontology, G-Prov, and used it to
enrich DMTO ontology, by associating provenance information with treatment
rules defined by their ontology. We have reused terminology from relevant ontolo-
gies such as DCT, Prov-O, BIBO. Specifically, DCT provides concepts that were
relevant to model parts of the reference information, the remaining information
was modeled using BIBO. Further, we represent the unstructured guideline data
as a structured RDF knowledge graph.
We build an application that comprises of different views rendered by results of
SPARQL queries triggered to the ontology and the KG. The different views of
the application represent the various use cases that our ontology-enabled system
can be used in. Our application serves as a proof of concept for the development
of systems that can benefit from identifying provenance of CDS treatment sug-
gestions.
Through, G-Prov we are taking a step towards enhancing transparency in health-
care by providing more information to healthcare practitioners regarding the
granularities of guideline evidence behind treatment suggestions. We are focusing
on improving the scalability and automation aspects of our system. Specifically,
we are exploring the use of NLP techniques to automatically/semi-automatically
identify evidence sentences that support a given guideline recommendation. Al-
though the G-Prov ontology has currently been used to model the ADA CPG, we
envisage that we can extend upon the ontology and introduce classes to model
guideline provenance from other CPGs.
7 Conclusion
In this paper, we have described the guideline provenance ontology: G-Prov,
which is a simple ontology for modeling provenance information of treatment
rules generated by CDS systems. In the ontology design process, we have reused
terms from several widely accepted ontologies to increase interoperability. We
have built a prototype application supported by the G-Prov ontology to show
its various use cases. Currently, we only map provenance of treatment rules to
the ADA 2018 CPG. However, G-Prov can be used to track the provenance
of information in other guidelines. We believe G-Prov is a good complement to
existing provenance ontologies, and thus it is our belief that the our paper will be
a useful contribution to both healthcare practitioners and CDS system designers
who are looking to develop treatment suggestion tools.
8 Acknowledgements
This work is partially supported by IBM Research AI through the AI Horizons
Network. We thank our colleagues from IBM (Ching-Hua Chen) and RPI (James
Hendler, John Erickson, Kristen Bennett, Rebecca Cowan) who provided insight
and expertise that greatly assisted the research.
74
8 N. Agu et al.
References
1. Chari, S., Qi, M., Agu, N.N., Seneviratne, O., McCusker, J.P., Bennett, K.P.,
Das, A.K., McGuinness, D.L.: Making study populations visible through knowl-
edge graphs. In: International Semantic Web Conference (ISWC). p. to appear.
Auckland, New Zealand (2019)
2. Dabrowski, M., Synak, M., Kruk, S.R.: Bibliographic ontology. In: Semantic digital
libraries, pp. 103–122. Springer (2009)
3. Deora, V., Contes, A., Rana, O.F., Rajbhandari, S., Wootten, I., Tamas, K., Varga,
L.Z.: Navigating provenance information for distributed healthcare management.
In: Proceedings of the 2006 IEEE/WIC/ACM International Conference on Web
Intelligence. pp. 859–865. IEEE Computer Society (2006)
4. El-Sappagh, S., Kwak, D., Ali, F., Kwak, K.S.: Dmto: a realistic ontology for
standard diabetes mellitus treatment. Journal of biomedical semantics 9(1), 8
(2018)
5. Horrocks, I., Patel-Schneider, P.F., Boley, H., Tabet, S., Grosof, B., Dean, M.,
et al.: Swrl: A semantic web rule language combining owl and ruleml. W3C Member
submission 21(79) (2004)
6. Kunze, J.A., Baker, T.: The dublin core metadata element set (2007)
7. Lebo, T., Sahoo, S., McGuinness, D., Belhajjame, K., Cheney, J., Corsar, D.,
Garijo, D., Soiland-Reyes, S., Zednik, S., Zhao, J.: Prov-o: The prov ontology.
W3C recommendation 30 (2013)
8. McCusker, J.P., Chastain, K., Rashid, S., Norris, S., McGuinness, D.L.: Setlr: the
semantic extract, transform, and load-r. PeerJ Preprints 6, e26476v1 (2018)
9. Miller, E.: An introduction to the resource description framework. Bulletin of the
American Society for Information Science and Technology 25(1), 15–19 (1998)
10. Noy, N.F., Fergerson, R.W., Musen, M.A.: The knowledge model of protege-2000:
Combining interoperability and flexibility. In: International Conference on Knowl-
edge Engineering and Knowledge Management. pp. 17–32. Springer (2000)
11. Sahoo, S.S., Bodenreider, O., Hitzler, P., Sheth, A., Thirunarayan, K.: Provenance
context entity (pace): Scalable provenance tracking for scientific rdf data. In: In-
ternational Conference on Scientific and Statistical Database Management. pp.
461–470. Springer (2010)
12. Valdez, J., Kim, M., Rueschman, M., Socrates, V., Redline, S., Sahoo, S.S.: Prov-
care semantic provenance knowledgebase: evaluating scientific reproducibility of
research studies. In: AMIA Annual Symposium Proceedings. vol. 2017, p. 1705.
American Medical Informatics Association (2017)
13. Zamborlini, V., Hoekstra, R., Da Silveira, M., Pruski, C., ten Teije, A., van Harme-
len, F., et al.: Generalizing the detection of internal and external interactions in
clinical guidelines. In: HEALTHINF. pp. 105–116 (2016)
75