Ontology for Risks in Medicines Supply Chain: Case of Public Hospitals in Morocco

For the supply system of medicines in public hospitals in Morocco, ontology is a solution that will improve understanding between actors, promoting information dissemination and their exploitation and promote a new approach to designing information systems. However, implementation of such an approach requires close collaboration between the various partners and requires an intensive exchange of data between them. These technical data must be transmitted, interpreted, managed, reused and stored in a consistent and standardized way. In this work, we developed an ontology using the OWLGRed tool from a UML model whose objective is risk management in the medicines supply chain in Morocco, principally in the public hospitals that supply itself in a centralized way through the supply division which handles the purchase, storage and distribution of pharmaceuticals in these institutions.


Introduction
Pharmaceuticals save lives and improve health and play a vital role in many aspects of health care.For this, they should beings available in hospital establishments at all times in sufficient quantities, in appropriate form, with assured quality and adequate information and at an accessible price [1].
Each partner contributes to the availability of pharmaceutical products at the hospitals, needs access to a set of information and knowledge distributed in several information systems (internal or external to its organization) to make effective choices while taking into consideration all the required constraints and risks that may disrupt the smooth functioning of the chain.The partners also need to share decisions and data with other partners that need it.However, each partner in the pharmaceuticals supply chain has its own domain of expertise and each speaks its own language.
With the emergence of ontologies as a new paradigm of information modeling, sharing and exchange of data between the different actors of a supply chain has become possible.Thus, a new integration and interoperability track has become promising.In effect, ontologies allow not only to exchange information between systems without loss of semantics but they also offer the ability to infer new information from existing ones [2].
Our aim in this paper is to use ontologies as integrator element and as efficient support for the identification of risks related to the medicines supply chain at the hospital centers in Morocco (excluding Universities Hospital Centers) to facilitate management and make explicit the common understanding of risks between the different actors of this chain.
To do this, in the 2nd section, we present a literature review of risks associated with the pharmaceuticals supply chain.Then, in the third section, we define some concept related to ontologies.The fourth section will include an analysis of the various risks associated with the medicines supply in public hospitals in Morocco and a description of the ontology developed for manage these risks.Finally, in Section 5 we conclude and present some future works.

Risks in pharmaceuticals supply chain in Morocco
The pharmaceuticals supply chain is defined as a set of processes, operations and organizations involved in the discovery, development and manufacturing of pharmaceuticals, which aims to provide medicines with the good quality at the right time and the right place and customers with optimal cot to be consistent with the objectives of the health system [3].Then, there are risks in a supply chain when unexpected events might disrupt the flow of materials on their journey from initial suppliers through to final customers.
Several studies propose a risk classification and modelization through medicines supply chain ( [4] and [5]).The majority of identified risks are similar to those prevailing in industrial supply chains despite the pharmaceutical products particularities.In Morocco, the medicines supply chain (Figure 1) is characterized by a context with large disturbances, a high uncertainty in supplying of medicine and medical devices and particularly by the significant risks which can decrease the coverage of patients needs and generate dangerous consequences for the Health Ministry, the patient, the hospital (reputation, decrease in performance) and for the care team.
In effect, the CESE report in 2013 emphasized that in level the hospital establishments (HE), the needs determination for medicines for the year n+1 is made on the basis of the average consumption of the previous years and not from the real needs and medical prescriptions.This low involvement of prescribers at the time of the needs expression generally produces overestimates and contributes to the accumulation of the medicines stocks and to their expiry.
Also, the CCMPS report (2012) stated that the medicines storage poses problems, with large quantities of expired medicines, for delays in deliveries, rupture of frequent stocks and the insufficient of the logistical means at central and local level.The medicines distribution suffers from a lack of coordination within the distribution network and the absence of an integrated information system for control the activities in medicines supply chain.In addition, the supply division (SD) as responsible for the consolidation of requirements, the tenders' launch, the reception, the storage and distribution of pharmaceutical products, the administrative management as well as the supply monitoring, doesn't have an information system to allow for effective traceability between the various links in the chain which makes the access to the information difficult and generates a set of risk at all levels [6].However, several efforts were made to prevent and solve these problems, mainly: the increase in medicines budget by the Health Ministry, the establishment of regional pharmacies, standardization and upgrading of hospital pharmacies and adapting and improving methods of the hospital pharmacy management [7].And despite all these efforts, the needs coverage of the population in medicines remains insufficient since the patient still complains of the non-availability of these products.So it isn't a question of meeting standards, procedures or standards of good practice, but to identify important concepts in the field, clarifying the relationships between objects to allow standardization of languages between partners.

Domain ontology for risks management in the medicines supply chain in Morocco
Domain ontology consists of a shared vocabulary between different partners of the medicines supply chain in Morocco.A first job is to build a corpus, which includes the definition and classification of risk and the relationship between partners.From the vocabulary contained in the corpus, we will extract the constituent primitive concepts of this ontology.

Definition of Ontology
The ontology is a new modeling paradigm of knowledge borrowed from philosophy and developed as part of the semantic web.They define the concepts, relationships between concepts, choices and constraints that must be respected.In effect, an ontology defines an organized whole (often in the form of taxonomy or semantic network) of usable concepts for formulating knowledge.[8].Ontologies specify explicitly the conceptual knowledge using a formal or semi-formal language.In the community of Knowledge Engineering, the Ontology term is often associated with a meta-model that describes the contents of a database, its properties, how it can be used and the vocabulary and syntax provided by the representation language.

Types and construction methods of ontology
There are several types of ontologies and its applications are diverse.We suggest in this context the classification proposed by [9] which proposes four types of ontologies: x Generic ontologies or upper ontologies : are an independent conceptualization of a problem or a particular area, for example: space, time, objective and event x Domain ontologies: describe the vocabulary related to an area such as health, industry and education.Also, it can describe the vocabulary associated with a task such as planning, diagnosis or purchase.This type of ontology defines the necessary knowledge to solving a particular type of work.
x Application ontologies: provide concepts based on a specific task and a particular area.x Representation ontologies: specify conceptualizations that underlie knowledge representation formalisms.In the literature, there are several works that have proposed the construction methods of these ontologies including the work of ( [10][11][12][13].The following table (Table 1) summarizes the definitions of these methods.

Methods Definitions
"Enterprise Ontology" [11] It is a method which aims: to improve communication between human beings, provide a basis for the specification of user applications and support interoperability.

"TOVE" [14]
A method that necessary to construct of a logical model of knowledge.
METHONTOL OGY [15] This method is developed by the Artificial Intelligence Laboratory for building ontologies in knowledge level.La TERMINAE [16] This method proposes the construction of ontologies from texts.

Unified process for ontology building
This method based on the Unified Process (UP) and UML.UPON is guided by diagrams of use because it aims to build useroriented ontologies.

Ontology Development Tools
There are various ontology languages like XML, RDF(S), DAML+OIL and OWL.Many ontology tools have been developed for implementing metadata of ontology using these languages as: x The Ontolingua server [17] located at Stanford University allows a user or group of users to view existing ontologies and cooperatively build new ontologies.
x WebOnto [18] developed at Knowledge Media Institute at the Open University.It is a web-based tool and mainly graphic to build ontologies cooperatively.
x ProtégéWin [19] was designed for the Department of Informatics Medical University of Stanford, especially for building ontologies.
x ODE [20] ODE (Ontology Design Environment) is a construction support tool for ontology in "knowledge level", which is independent of any formal language.
x OWLGred [21] allows creating, editing and viewing ontology.It offers a comprehensive overview of OWL based on UML.We aim in this work to use domain ontologies as efficient support risk management in the medicines supply chain in Morocco to facilitate integration, internal and external, of the different partners in this chain.The developed ontology will be built using a UML model.Indeed, we have interest in using the "Unified Process for ontology building" method to identify different risk which may prevent access to medicines and also describe the supply process and partners of this chain.OWLGred tool was chosen to automatically pass our conceptual model expressed as a UML class diagram to a formal semantic ontology represented by OWL.

Conception of the domain ontology
The domain ontology (DO), as was mentioned before, defines all concepts related to a domain of activities semantically, it concerns here the risk at the level of the medicines supply chain in Morocco.It aims primarily to "assemble and unify the terms related to risk management".Also, it provides a vocabulary of concepts used in the domain with their definition, similar concepts, related concepts, etc.The conception of this ontology was made in the Unified Modeling Language (UML).It is a simple formalism for experts in the domain that facilitates the process of the ontology constitution.
UML profiles used to specialize a UML model.They can be used in all models without modifying their structure.Moreover, it should be noted that such reasoning, although it is possible, isn't exploitable by a machine because it isn't an automatic reasoning since UML profiles don't allow any form of automatic semantic processing.In effect, the domain ontology modeling in UML must then turn into formal ontology.It can then be edited in protected and enriched environment.Note that several research papers suggest modeling ontologies with UML, including [22].In [23], the authors develop an approach which aims the visualization of OWL ontologies using the UML graphical notation.To create an ontology with UML (enriched) and then automatically switch to OWL (or vice versa), [24] suggest OWLGred tool.We chose to use the same tool to automatically switch from a conceptual model expressed as a UML class diagram to a formal semantic ontology represented by OWL.
The following figure (Figure 2) shows partners of the medicines supply chain in Morocco (Supply division, suppliers, warehouses, delegations and public hospitals) where each rectangle represents a class, and lines with a triangle represent the generalization relationship between classes.
Identifying needs in hospitals is often left to head nurses or doctors who transfer the patient needs to the service needs (Need Class) and transmit them to the hospital pharmacy (Hospital Pharmacy Class).The Need Class also includes the programmed and nonprogrammed (urgent) needs for one or more services (Service Class).All these needs are then transferred to the Regional Delegations and Provincial Directorates (Delegation Class) that compile annual commands on pharmaceuticals of one or more hospitals (excluding the CHU : autonomous supply).Then, the commands elaborated send to the SD (SD Class) to begin the process of purchasing from suppliers.The global distribution is to deliver the products to the service (Service Class) in advance on presentation of a weekly order for ordinary commands for a period of a week or an order for pharmaceutical complement which allows the service to order the products which don't available in its stock at any time.The products are delivered just after the deposit of a delivery order signed by the service head.These products are then stored at the service and issued to healthcare personnel (Nurse Class) and (Doctor Class) who administer by a medical prescription.Concerning the nominative distribution, allows the products delivery to a patient (Patient class) and not a service which plays an intermediary role between the pharmacy and the patient, on presentation of a prescription to the nominative delivery of products.Also, the proposed domain ontology allows different partners to identify any risk can be presented in the medicines supply chain into different categories, subcategories and management level (operational, tactical and strategic) (Figure 3).

Conclusion
In this article we proposed domain ontology as a decision support system for the risks management in medicines supply system at the public hospitals of Morocco.The proposed system provides three functions: identifying risks in the medicines supply chain at the level of the public hospitals (except University Hospital Center), semantic description of each risk category and risk management by the various partners in the chain.
The decision support system proposed, will be validated in a larger context with the final users and also the impact of the evolution of ontologies should be studied on this system.

Fig. 1 .
Fig. 1.Medicines supply process for public hospitals in Morocco.
pharmacy, the products delivered by the central warehouses (Warehouse Class) are then stored in the stock of pharmacy (Stock Class) and distributed to various services according to a schedule adapted for each hospital.The dispensation is made according to the care units type (Service Class), product type (PH Class) and resources available but generally we can distinguish between two main modalities: Global and Nominative.Indeed, the Distribution Class creates a flexibility of choice of the distribution methods (Global and Nominative Classes) of medicines and medical device (Medicines and MD Classes). hospital