=Paper= {{Paper |id=None |storemode=property |title=Pharmaceutical Validation of Medication Orders Using an OWL Ontology and Business Rules |pdfUrl=https://ceur-ws.org/Vol-930/p5.pdf |volume=Vol-930 |dblpUrl=https://dblp.org/rec/conf/semweb/ChnitiBDAC12 }} ==Pharmaceutical Validation of Medication Orders Using an OWL Ontology and Business Rules== https://ceur-ws.org/Vol-930/p5.pdf
Pharmaceutical Validation of Medication Orders
 Using an OWL Ontology and Business Rules

Amina Chniti1,2 , Abdelali BOUSSADI2,4,5,6 , Patrice DEGOULET2,4 , Patrick
                           Albert1 , Jean Charlet2,3
                                  1
                                    CAS France, IBM
                        {amina.chniti,albertpa}@fr.ibm.com
    2
      Inserm UMRS 872 q.20, Ingénierie des Connaissances en Santé, Paris, France
                                Jean.Charlet@upmc.fr
                                3
                                  AP-HP, Paris, France
                4
                   Hôpital Européen George Pompidou, Paris, France
              {abdelali.boussadi,patrice.degoulet}@egp.aphp.fr
                      5
                        UPMC Université Paris 06, Paris, France
                 6
                   Université René Descartes Paris 05, Paris, France



        Abstract. Ontologies are description of domains encoded in a formal
        language while Business Rules are description of business policies en-
        coded in a natural controlled language. In this paper we present an ap-
        plication of pharmaceutical validation of medication order based on an
        OWL ontology and business rules or more specifically clinical decision
        rules. This application has been developed based on a prototype that en-
        ables business users to author, in a controlled natural language, execute
        and manage their Business Rules over OWL Ontologies.
        Keywords: OWL Ontology, Business Rule, Clinical Decision Rule, Phar-
        maceutical Validation.


1     Introduction
Ontologies are more and more used to model the business knowledge which is due
to their power of expressiveness and to their flexibility. On the other hand, many
business applications are nowadays built based on Business Rules especially after
the emergence of the BRMS (Business Rules Management System). Business
Rules are a description of a business policy, encoded in a natural controlled
language. They define or specify constraints of some aspect of the business7 and
enable automating business decisions. An example of a business rule is given in
the following :
      IF the presentation name of the drug is ”GLUCOPHAGE 850MG TAB”
      and the dosage unit of the dosage regimen phase is ”TABLET”
      THEN the prescripion is not valid;

   In this paper, we present an application of pharmaceutical validation that
enable to automate the decision of validation of medication orders. It is based on
7
    http://www.businessrulesgroup.org
an OWL ontology and business rules, or more specifically clinical decision rules.
The OWL 8 (Web Ontology Language) ontology models the most pertinent
entities (concepts and properties) of pharmaceutical validation activity used in
the Hôpital Européen Georges-Pompidou (HEGP) (Georges Pompidou European
Hospital) 9 . The rules, test on the values given to the entities described in the
ontology and assert if a medication order is valid or not [2].
    The fact of using business rules enables the business user (i.e.pharmacist,
physician) to be involved in the implementation of the application as he/she can
author the rules in a natural controlled language.
    Business rules and ontology have already been combined to support clinical
decision [6] [7]. However, end user involvement in the design and the implementa-
tion of the application is a neglected aspect. In this study we propose to involve
the end user (pharmacists, physicians, nurses) in the implementation of the ap-
plication and to experiment the business rules designed as a clinical decision
rules.
    To develop this application, we first implement a prototype 10 , OWL plug-in
for WODM, that enable authoring and executing business rules over OWL on-
tologies [3]. For this, we based on the infrastructure offered by the Business Rule
Management System (BRMS) WebSphere Operational Decision Management
(WODM) 11 and added as input OWL ontologies.
    This paper is organized as follows; Section 2 present the OWL plug-in for
WODM. Then Section 3 describes the application of pharmaceutical validation
of medication order. Finally Section 4 concludes and presents our perspectives.


2   Proposed Approach

WODM offers an infrastructure that enables business users to author, - in a con-
trolled natural language -, execute and manage business rules in a collaborative
way. As the majority of BRMS, it uses an object oriented models to formalize the
domain knowledge. In WODM, this object oriented model is called BOM (Busi-
ness Object Model). The BOM represents the entities of a given business (i.e.
patient, age). It is generated over the XOM (eXecutable Object Model) then ver-
balized. The XOM is the model enabling the execution of rules. It references the
application objects and data, and is the base implementation of the BOM. The
XOM can be built from compiled Java classes (Java execution object model) or
XML Schema (dynamic execution object model). The verbalization of the BOM
consists of generating a controlled natural language vocabulary (VOC) which
enables authoring the business rules (i.e. the patient, the age of the patient).
8
   http://www.w3.org/2004/OWL/
9
   http://www.aphp.fr
10
   This work is partially founded by the European Commission under the project ON-
   TORULE (IST-2009-231875).
11
   http://www-01.ibm.com/software/decision-management/operational-decision-
   management/websphere-operational-decision-management/
2.1    Authoring Business Rules over OWL Ontologies

To enable business users to author business rules, in a natural controlled lan-
guage, we developed the WODM OWL plug-in. This plug-in exploits infrastruc-
ture offered by WODM to import OWL ontologies within it. The main com-
ponent for authoring rules in WODM is the BOM. For this, we performed a
mapping of OWL concepts (TBox) into the BOM. Thus, when we import an
OWL ontology within WODM, the BOM is automatically generated and the
functionalities offered by the BRMS can be used. The general idea of the map-
ping is: ontology concepts are mapped into BOM classes and the properties are
mapped into attributes of the classes. Nevertheless, due to the difference of the
power of expressiveness between OWL ontology and the BOM, there are some
OWL construct that could not map into the BOM [3].


2.2   Executing Business Rules over OWL ontologies

The process of executing business rules in WODM consists of several steps.
Business rules, authored in a controlled natural language are translated into ex-
ecutable rules, which are written in a formal technical rule language IRL (ILOG
Rule Language). During this translation, the references to the BOM’s classes and
properties are translated to references into the XOM. When the input provided
to WODM is a Java object model, the XOM is built from this model. But in our
case, the input provided to WODM is an OWL model.
    To execute business rules authored over ontologies, we perform a second map-
ping of OWL/BOM entities to a XOM using Jena. Jena is a Java framework,
including an ontology API for handling OWL ontologies, which allows gener-
ating Java objects from the entities of the ontology. These Java objects then
constitute the XOM. The use of Jena provides an execution layer for the OWL
ontologies. This execution layer provides inference mechanisms on this model
and the mapping of OWL concepts, properties, and individuals to a Java object
model.


3     Experimentation

The method described above enables to author and execute business rules over
OWL ontology. This method can be used in different business domain. In our
case, to experiment our work, we used a pharmaceutical validation use case to
implement clinical decision rules for pharmaceutical validation of medication
orders.
   To illustrate our work, we present a business scenario that stages three per-
sonas representing business users involved in the design and the implementation
of a set of clinical decision rules for pharmaceutical validation of medication
orders (see Figure 1).
   Marc is the business analyst. His mission is to formalize the business knowl-
edge and to make sure that the business model (i.e. ontology) is correct, complete
                             Fig. 1. Business scenario.




and valid. Alice is the domain expert. In this business scenario two domain ex-
perts interact which are a pharmacist and a physician. They understand the
formalization of the rules and uses business rules tools. They are in charge of
editing the clinical decision rules in the business application. Joana is the oper-
ational user who is the pharmacist. She uses the application to verify if a given
prescription is valid or not.
    The ontology used to build this business application and the authored busi-
ness rules that we will show in the following have been built based on the work
presented in [2]. This work has been made in collaboration with pharmacists
and physicians from the Georges Pompidou University Hospital [4]. The ontol-
ogy is composed of 17 concepts and 25 properties. We will focus on 5 concepts
and their properties used to author the rules we present in this paper. The on-
tology contains a concept Patient which has LabResult and is concerned by
a Prescription that has a DosageRegimenPhase. The rules authored over
this ontology test on the presentation name of a Drug, the dosage unit and
the dosage of the DosageRegimenPhase of a Prescription and on the GFR
(Glomerular Filtration Rate) of the LabResult of a Patient. Depending on the
values given to these properties, they assign if a Prescription is valid or not.
    When Marc finish the edition of the ontology using an ontology editor (i.e.
Protégé), he imports it into WODM which automatically generate the BOM.
Once the BOM is generated, the domain experts (pharmacists and physicians)
author the clinical decision rules in natural controlled language. Two examples
of authored rules are presented in the following (see Figure 2 & 3).
   The rule in Figure 2 tests if :
 – the presentation name of a drug is “GLUCOPHAGE 1000MG TAB” or
   “GLUCOPHAGE 1000MG CPR COATED”
 – the dosage unit of the dosage regimen phase is “TABLET”
 – the dosage of the dosage regimen phase is more than 1
 – the GFR of the lab result of a patient is more than 50
Then it sets that the prescription is not valid.




                    Fig. 2. GLUCOPHAGE (Metformin)-1000.



   The rule in Figure 3 tests if:
 – the presentation name of a drug is “GLUCOPHAGE 850MG TAB” or “GLU-
   COPHAGE 850MG CPR COATED”
 – the dosage unit of the dosage regimen phase is “TABLET”
 – the dosage of the dosage regimen phase is more than 3
 – the GFR of the lab result of a patient is more than 80
Then it sets that the prescription is valid.
    The pharmacist enters the data concerning a prescription and launches the
execution of the rules which will determine if the prescription is valid or not. For
example, Joana enters data concerning two prescriptions given to two different
patients (see Table 1).
    Prescription 1 for patient 1 who has the GFR of his lab result equals to 90.
The dosage unit of the dosage regimen phase of the prescription is TABLET and
its dosage is 2. The prescription contains a drug called GLUCOPHAGE 1000MG
TAB. In this case the rule called GLUCOPHAGE - 1000 TABLET Rule (see
Figure 2) will be launched and set the validation of the prescription to false.
                    Fig. 3. GLUCOPHAGE (Metformin)- 850.




   Prescription 2 for patient 2 who has the GFR of his lab result equals to 95.
The dosage unit of the dosage regimen phase of this prescription is TABLET
and its dosage is 4. The prescription contains a drug called GLUCOPHAGE 850
MG TAB. In this case the rule called GLUCOPHAGE (Metformin) 850 (see
Figure 3) will be launched and the validation of the prescription will be set to
true.

                  Table 1. Prescriptions of the pharmacist

                Presentation name Dosage unit Dosage GFR
            P1 GLUCOPHAGE 1000MG TAB TABLET     2     90
            P2 GLUCOPHAGE 850 MG TAB TABLET     4     95




4   Conclusion

In this paper, we present an application of pharmaceutical validation of medi-
cation orders that implement clinical decision rules, in a natural controlled lan-
guage, over an OWL ontology. In order to develop this application we use the
prototype described in Section 2 that enables authoring and executing business
rules. The clinical decision rules designed in this study will be integrated with
the HEGP clinical information system as an alert system for more assessment.
    In perspective, we propose to improve the rule presented in this work in
order to have recommendations that enable to revise the invalid prescription.
Such recommendations, considered as alerts, inform the pharmacist about the
cause of the invalidity of a prescription.
    One particularity of ontologies is that they evolve over time. Ontology evo-
lutions consist of changes that could impact an ontology. The business rules
depend on the entities of the ontology and its evolution may have an impact
on them and causes inconsistencies. This is an issue on which we focus and for
which we developed the MDR approach (Model-Detect-Repair) [5]. This ap-
proach enables to tracks ontology changes, detects the rule inconsistencies that
could be caused by a change and then proposes solution, called repair, to repair
the inconsistencies.


References
1. G. Antoniou, C. V. Damasio, B. Grosof, I. Horrocks, M. Kifer, J. Maluszynski,
   and P. F. Patel-Schneider. Combining rules and ontologies: A survey. Tech-
   nical Report IST506779/Linkoping/I3-D3/D/PU/a1, Linkoping University, 2004.
   http://rewerse.net/publications/.
2. A. Boussadi, C. Bousquet, B. Sabatier, T. Caruba, P. Durieux, and P. Degoulet. A
   business rules design framework for a pharmaceutical validation and alert system.
   Methods of Information in Medicine, 2011.
3. A. Chniti, S. Dehors, P. Albert, and J. Charlet. Authoring business rules grounded in
   owl ontologies. In M. Dean et al. (Eds.), editor, RuleML 2010 : The 4th International
   Web Rule Symposium: Research Based and Industry Focused. LNCS 6403, Springer-
   Verlag Berlin Heidelberg 2010, 2010.
4. P. Degoulet, L. Marin, M. Lavril, C. Le Bozec, E. Delbecke, J-J. Meaux, and L. Rose.
   The hegp component-based clinical information system. . Int J Med Inform; 69(2-
   3):, pages 115–126, 2003.
5. M. Fink, A. El Ghali, A. Chniti, R. Korf, A. Schwichtenberg, F. Lévy, J. Pührer,
   and T. Eiter. D2.6 consistency maintenance. final report. ONTORULE Delivrable,
   http://ontorule-project.eu/deliverables., 2011.
6. V. Kashyap, A. Morales, and T. Hongsermeier. On implementing clinical decision
   support: Achieving scalability and maintainability by combining business rules and
   ontologies. AMIA Annu Symp Proc, pages 414–418, 2006.
7. Y. Kawazoe and K. Ohe. An ontology-based mediator of clinical information for
   decision support systems: a prototype of a clinical alert system for prescription.
   Methods Inf Med, vol. 47, no. 6, pages 549–559, 2008.