You are on page 1of 9

www.elsevier.

com/locate/procedia
Available online at www.sciencedirect.com

ScienceDirect
Available online at www.sciencedirect.com
Procedia Computer Science00 (2017) 000–000
CENTERIS - International Conference on ENTERprise Information Systems / ProjMAN -
www.elsevier.com/locate/procedia
ScienceDirect
International Conference on Project MANagement / HCist - International Conference on Health
and Social Care Information Systems and Technologies, CENTERIS / ProjMAN / HCist 2017, 8-10
Procedia Computer Science 121 (2017) 843–851
November 2017, Barcelona, Spain

Quality
CENTERISAssessment
- InternationalofConference
an Emergency CareInformation
on ENTERprise ProcessSystems
Model/ ProjMAN
based on -
International Conference onStatic and Dynamic Metrics
Project MANagement / HCist - International Conference on Health
and Social Care Information Systems and Technologies, CENTERIS / ProjMAN / HCist 2017, 8-10
November
a, 2017, Barcelona, Spain a,b,
Ghada Haouari *, Sonia Ayachi Ghannouchi *
QualityHigh
Assessment of an Emergency Care Process Model based on
Institute of Management of Sousse- Sousse University,Street Abdlaaziz il Behi. Bp 763, Sousse 4000, Tunisia
a

Laboratory RIADI-GDL-Manouba University, Tunisia


b

Static and Dynamic Metrics


Abstract Ghada Haouaria,*, Sonia Ayachi Ghannouchia,b,*
a
High Institute
Healthcare domain is a vitalof domain
Management of Sousse-
in which many Sousse University,Street
processes Abdlaaziz
are dealt with. More il Behi. Bp 763,inSousse
precisely, 4000, Tunisia
an emergency department, it is
b
Laboratory RIADI-GDL-Manouba University, Tunisia
fundamental to have an emergency care process with high quality. In this paper, we propose a solution based on an improved BPM
lifecycle for assessing and improving the quality of such a process, using quality metrics. These metrics are inspired from those of
software engineering and are classified into two main categories: static metrics and dynamic metrics. Static metrics are applied to
BPMN models and allow obtaining values related to complexity and coupling. Dynamic metrics concern the execution of BPMN
Abstract
models and obtain their values from log files. Our proposed approach has been applied for the emergency care process. For this
purpose, we evaluated its quality in two steps (static then dynamic) and finally some recommendations have been given in order to
Healthcare domain is a vital domain in which many processes are dealt with. More precisely, in an emergency department, it is
allow the designer to improve our emergency care process model.
fundamental to have an emergency care process with high quality. In this paper, we propose a solution based on an improved BPM
lifecycle for assessing and improving the quality of such a process, using quality metrics. These metrics are inspired from those of
© 2017 The Authors. Published by Elsevier B.V.
software engineering and are classified into two main categories: static metrics and dynamic metrics. Static metrics are applied to
Peer-review under responsibility of the scientific committee of the CENTERIS - International Conference on ENTERprise
BPMN models and allow obtaining values related to complexity and coupling. Dynamic metrics concern the execution of BPMN
Information Systems / ProjMAN - International Conference on Project MANagement / HCist - International Conference on
models and obtain their values from log files. Our proposed approach has been applied for the emergency care process. For this
Health and Social Care Information Systems and Technologies.
purpose, we evaluated its quality in two steps (static then dynamic) and finally some recommendations have been given in order to
allow the designer to improve our emergency care process model.
© 2017
© 2017 The
The Authors.
Authors. Published
Published by Elsevier B.V.
by Elsevier B.V.
Peer-review under responsibility of the scientific committee of the CENTERIS - International Conference on ENTERprise
Peer-review under responsibility of the scientific committee of the CENTERIS - International Conference on ENTERprise
Information Systems / ProjMAN - International Conference on Project MANagement / HCist - International Conference on
Information Systems
Health and Social / ProjMAN
Care - International
Information Systems andConference on Project MANagement / HCist - International Conference on
Technologies.
Health and Social Care Information Systems and Technologies.
* Corresponding author. Tel.: +216 73 332 979.
E-mail address:Ghada.Haouari@outlook.fr // s.ayachi@coselearn.org

1877-0509© 2017 The Authors. Published by Elsevier B.V.


Peer-review under responsibility ofthe scientific committee of the CENTERIS - International Conference on ENTERprise Information Systems /
ProjMAN - International Conference on Project MANagement / HCist - International Conference on Health and Social Care Information Systems
and Technologies.
* Corresponding author. Tel.: +216 73 332 979.
E-mail address:Ghada.Haouari@outlook.fr // s.ayachi@coselearn.org

1877-0509© 2017 The Authors. Published by Elsevier B.V.


Peer-review under responsibility ofthe scientific committee of the CENTERIS - International Conference on ENTERprise Information Systems /
ProjMAN - International Conference on Project MANagement / HCist - International Conference on Health and Social Care Information Systems
and Technologies.

1877-0509 © 2017 The Authors. Published by Elsevier B.V.


Peer-review under responsibility of the scientific committee of the CENTERIS - International Conference on ENTERprise Information
Systems / ProjMAN - International Conference on Project MANagement / HCist - International Conference on Health and Social
Care Information Systems and Technologies.
10.1016/j.procs.2017.11.109
844 Ghada Haouari et al. / Procedia Computer Science 121 (2017) 843–851
Ghada Haouari and Sonia Ayachi Ghannouchi / Procedia Computer Science00 (2017) 000–000

Keywords:Emergency Healthcare Process; BPM; BPMN Model; Quality metrics.

1. Introduction

The field of healthcare is a vital area where the change is needed and this is due to the emergence of technologies
that play a major role in the improvement and evolution of services. Indeed, the medical sector is very demanding in
NICT solutions (New Information and Communication Technologies) since they present a factor of improvement of
performance and efficiency. In addition, healthcare institutions are always looking for better satisfaction of their
clients by offering them better and more efficient services.
Consequently, the development of the technology offers real opportunities to answer to these challenges that this
sector meets and the BPM approach seems to be the right solution to offer improvements in order to ensure better
quality, which presents an essential element in the healthcare field.
Business Process Management (BPM) is a contemporary approach used to provide a complete description of the
company's processes and their interactions, in order to optimize and automate them.
BPM approach refers to a set of steps and, design (or modeling) represents a major step in the lifecycle of BPM.
Its objective is to create a representation that describes the functioning of the process as it is executed in a company.
Therefore, the assessment of the quality of process models is essential to improve them. For this, we propose in this
work to consider a set of quality metrics inspired from software engineering. These metrics will measure the quality
of the existing model and their values will be compared with threshold values. Whenever they deviate from these
threshold values, a list of recommendations will be left to the designer on the basis of which he/she elaborates the
model of the new process.
The main originality of this work concerns the fact that it is not limited to static metrics, but also considers dynamic
metrics that take their values from log files. In addition, this work proposes an improved model of the BPM lifecycle,
based on two types of quality improvements: the first one is based on static metrics and the second one is based on
dynamic metrics. In this context, our contribution in this paper is to apply a new solution for the assessment of quality
for an emergency care process model, modeled in Business Process Model and Notation (BPMN).
The remainder of this paper is organized as follows: In section 2, we present the considered emergency healthcare
process. In section 3, we present our proposed approach which is validated by the creation of a prototype and has been
tested with the emergency care process. In section 4, we overview quality metrics defined in the business process
domain. Our solution is presented in section 5.We give an overview of related works in section 6. Finally, we conclude
this paper with a summary of the presented work and an outline of its future works.

2. Considered emergency healthcare process

Our chosen care process is that of the emergency department of Farhat Hached Hospital of Sousse modeled and
presented by2.For our work, we chose to work on the model of valid patient case (see Fig. 1). A valid patient is a
patient who can walk alone between different areas of the emergency department. The actors involved in this process
are: the administrative agents, the physicians, senior technicians and nurses and also patients.
The functioning of this process is as follows:
 The care process is started by the simple arrival of a patient and his presence in front of the registration desk.
 According to his health status, the patient is going to pass through several stages:
 Consultation: this is the most important task. It is assumed by doctors and interns.
 Complementary examinations: these examinations can be realized in the department itself (e.g. blood
samples), as they may require the displacement of the patient to another department (e.g. scanner).
 Treatment and care: they are realized by doctors or nurses in the surgery box.
 Special consultation: the patient is sent to a special consultation through a letter written by the emergency
physician.
 Hospitalization: the doctor is responsible for making decisions about the hospitalization of the patient if his
situation requires long treatments. It is made in observation rooms.
Ghada Haouari et al. / Procedia Computer Science 121 (2017) 843–851 845
Ghada Haouari and Sonia Ayachi Ghannouchi / Procedia Computer Science00 (2017) 000–000

 Finally, the patient comes to the cash box in order to accomplish the task of payment of his/her care costs. Then,

the process ends when the patient leaves the emergency department.
Our care process is modeled, implemented and executed using the tools provided by the Business Process
Management System (BPMS) Bonita.

3. Proposed approach

Emergency is an area in which change must occur. Indeed, a better satisfaction of the patient and a service of
quality are necessary. The process of care is the basic process in the emergency department, in which several actors
intervene to ensure patient satisfaction. In this context, it seems so important to model the process and make the
necessary changes in order to improve it and to ensure a good quality of service.
The objective of our research is to propose a guidance approach for improving the quality of our emergency care
process, more precisely the quality of the BPMN corresponding model. Therefore our work consists in evaluating the
quality of this BPMN model at the modeling (design) phase as well as at the execution phase. It also consists in
generating a set of recommendations for designers in order to improve the quality of their emergency process model.
For this purpose, we adopt an extension of the BPM lifecycle, based on its main phases and adapted to our approach
(see Fig. 2 page 6).In this extension of BPM lifecycle, we have proposed to add the following phases:
 Design-based Evaluation of the model: in this phase, the elaborated model is evaluated on the basis of a set of
static quality metrics that are taken from literature (presented in the following section: subsection 4.1).In our
approach, we focused only on the use of the following static metrics: NOA, NOAJS, CFC, ICP and ECP. In fact,
these are the metrics for which we will be able to determine their values for the process and the BPMS that will
be considered.
 Generation of recommendations (1): this step allows the user - who is the modeling expert - to improve the
quality of his model by referring to the generated recommendations. These recommendations concern the quality

Fig. 1.Emergency care process model.


846 Ghada Haouari et al. / Procedia Computer Science 121 (2017) 843–851
Ghada Haouari and Sonia Ayachi Ghannouchi / Procedia Computer Science00 (2017) 000–000

of the model itself according to static point of view. They appear each time a static metric takes an inappropriate
value.
 Execution-based Evaluation of the model: at this stage, the quality of the process is evaluated on the basis of
its execution using dynamic quality metrics which are inspired by those applicable in the software domain
(presented in the following section: subsection 4.2).In our approach, we focused only on the use of the following
dynamic metrics: ED(p), ED(t) and WT(t’).
Note: The evaluation of the model after the execution phase is based on the analysis of the events. Therefore, we
are going to use the ".log" files produced by the BPMS Bonita. In the processed case, we proposed to combine
the two metrics ED(t) and WT(t ') to have the metric ET[T , T ']. In fact only the dates of the end of task execution
are registered in Bonita.
 Generation of recommendations (2):this phase allows the user - who is the business expert - to make the
necessary modifications to improve the quality of his process by referring to the generated recommendations.
These recommendations can bring us in certain cases to the initial model to improve it, as in some other cases,
they can indicate and help to solve defects resulting from the implementation phase. This is illustrated by the
return arrows presented in the lifecycle.

4. Quality metrics

In the following part, we will present the selected quality metrics used in our approach for the evaluation of quality
of emergency care process model. These metrics, selected in our work, are classified into static and dynamic metrics.

4.1. Static metrics

Static metrics are used to evaluate the quality of the model after its elaboration (at the modeling phase of the BPM
lifecycle).
The metrics NOA (Number of Activities)4 and NOAJS (Number of Activities, Joins and Splits) 4 can be used to
measure the complexity of business process models. NOA counts the number of activities and NOAJS counts the
number of activities and the number of "joins" and "splits" of a process.
The metric CFC (Control-Flow Complexity)4evaluates the complexity of the process introduced by gateways such
as "XOR-Split", "OR-Split" and "AND-Split". The main idea of this metric is to evaluate the number of scenarios that
can be considered during the execution of the process.

CFC   CFC XOR  Split (a)   CFC ORSplit (a)   CFC ANDSplit (a)
(1)
Where: a is an activity and n is the number of transitions going out of an activity

CFCXOR-Split (a) = n, CFCOR-Split (a) = 2n-1, CFCAND-Split (a) = 1

The metrics ICP (Imported Coupling of a Process)7 and ECP (Exported Coupling of a Process)7are used to
determine the degree of coupling based on the number of messages / sequence flows in a business process model. ICP
counts, for each (sub-) process, the number of messages / sequence flows sent by either the tasks of the (sub) process
or the (sub-) process itself. ECP counts, for each (sub-) process, the number of message/sequence flows received by
either the tasks of the (sub-) process or the (sub-) process itself.

4.2. Dynamic metrics

This type of metric is used to evaluate the quality of the model after the execution of the process (at the execution
phase of the BPM lifecycle).In the following, we present the group of metrics that we have adapted from the software
metrics presented in11.
Ghada Haouari et al. / Procedia Computer Science 121 (2017) 843–851 847
Ghada Haouari and Sonia Ayachi Ghannouchi / Procedia Computer Science00 (2017) 000–000

 ED(p) metric (Execution Duration of a process):This metric is used to measure the execution duration of a
process.

ED(p) = date of the end of process execution – date of the start of process execution (2)

 ED(t) metric (Execution Duration of a task):This metric is used to measure the execution duration of a task.

ED(t) = date of the end of task execution – date of the start of task execution (3)

 WT(t’) metric (Waiting Time of task execution):This metric is used to measure the waiting time of the
execution of a task.

WT(t’) = date of the start of the execution of task t’ – date of the end of the execution of task t (4)
(t is the task that precedes the task t’)

 ET[T, T '] metric (Elapsed Time [T, T ']): This metric counts the elapsed time between two tasks T and T 'of a
process instance (going from the end of the task T - the previous task - to the end of the task T').

 GWT metric (Global Waiting Time): This is the time deducted from the sum of the waiting times of the
execution of all tasks in a process instance.

GWT  WT (t ' ) (5)

 OCC metric: This metric is used to count the number of occurrences of a particular task for a given instance of
a process.

5. Developed prototype

In order to illustrate our proposed solution, we developed a prototype. It is a simple web application implemented
by HTML and PHP. This prototype allows to measure the level of quality of the emergency care process model.
As mentioned in the section 3, the adopted BPM lifecycle model contains additional steps to evaluate the model
after the modeling phase as well as after the execution phase. These additional steps involve two types of users:
modeling expert and business expert. As a result, we have dedicated in our application two spaces: Modeling space
and Execution space. We are going to present below our prototype by considering the case of the emergency care
process.

5.1. Modeling space

In this space, the modeling expert can evaluate the quality of his/her model after its elaboration. Firstly, the user
will find an interface called “The quality of the model” in which there will be displayed a table summarizing the set
of quality metrics according to which the model will be evaluated (the metrics: NOA, NOAJS, CFC, ICP and ECP).
After having an idea about the used quality metrics: their types and their descriptions, the modeling expert has to
assign a value to each of these metrics. These values represent the optimal values, according to the expert, which will
be considered as threshold values for evaluating the model. Secondly, and by clicking on the 'validate' button, another
interface named 'Evaluation' will be presented. The modeling expert will obtain in this interface (see Fig. 3) details
concerning the analysis of the process model on the basis of the considered static metrics, as well as some
recommendations allowing him/her to improve the quality of the model.
These recommendations will be indicated whenever one of the considered metrics does not have an acceptable
value. They appear for example as follows: “The NOAJS metric exceeds the threshold value. Please reduce the
number of tasks or the number of gateways.”
848 Ghada Haouari et al. / Procedia Computer Science 121 (2017) 843–851
Ghada Haouari and Sonia Ayachi Ghannouchi / Procedia Computer Science00 (2017) 000–000

5.2. Execution space

In this space, the business expert can evaluate the quality of the process on the basis of its execution and make the
necessary modifications by referring to the generated recommendations. This is done by consulting the two interfaces:
the “Recommendations” interface and the “Details” interface. Figure 4 shows that in order to improve the quality of
the care process, it is advised to follow the proposed recommendations.

Fig. 2.Adopted lifecycle of BPM. Fig. 3. Details of the process analysis obtained based on its model.

These recommendations are related to the task “Treatment” which has exceeded the threshold value. In order to better
understand the causes of this problem and to find the way to remedy it, the user can consult the details as shown in
figures 5 and 6. Figure 6 (page8) indicates that according to quality metrics, 5 instances out of 7 demonstrated an
overtaking of threshold value related to the task “Treatment” (ET[5]).
Indeed, at the “Details” interface, the user can get an idea of the set of quality metrics used to evaluate the quality
of the process on the basis of its execution: ED(p) metric and ET [T , T '] metric. Also, we gave to the user a table
summarizing the existing tasks in the process by associating them with the values generally taken in normal conditions
of functioning and which are considered as threshold values. These values of the metric ET[T, T '] are attributed by a
business expert (see Fig. 5).After getting an idea about the set of used metrics, the business expert can detect which
tasks have higher values than threshold. Such a situation is presented by a different colour. As illustrated in figure 6,
we note that each of the following tasks in the care process: “Consultation”(ET[2]), “Doctor's opinion after
complementary examinations”(ET[4]), “Treatment” (ET[5]) and “Doctor’s opinion after treatment” (ET[6]) were in
a case of an overtaking of the threshold value. The values of the metrics presented in figure 6 are taken from the result
of the automatic analysis of the ".log" file produced by Bonita. This log file is obtained through the execution of seven
instances of our process. These executions have been fulfilled according to real instances collected by observations of
real valid patients in the emergency department of Farhat Hached Hospital of Sousse. These observations were
simulated after adapting them to our chosen "to-be" model.
According to the details of the process analysis based on its execution using the quality metrics, a list of
recommendations was displayed to help the business expert improving the quality of the process. For example, among
the proposed recommendations for improving our emergency care process: “For the task Consultation we propose
to: Increase the number of consultation boxes or Increase the number of doctors.”
The display of the recommendations list depends on the set of tasks that exceed the threshold value and also on the
number of realized instances. In other words, if a task marks an exceeding of its threshold value for the half of the
number of realized instances (or more), the list of recommendations related to this task will be presented. For this
reason, figure 4 presents only the recommendations related to the task “Treatment”, although some other tasks have
exceeded their threshold values according to the details of the analysis.
Ghada Haouari et al. / Procedia Computer Science 121 (2017) 843–851 849
Ghada Haouari and Sonia Ayachi Ghannouchi / Procedia Computer Science00 (2017) 000–000

5.3. Learning possibilities

The “Recommendations” interface not only gives the user an idea of the proposed recommendations for improving
the quality of the process, but also allows the user to add

a new recommendation (see Fig. 4).

This recommendation will be saved and taken into account in the next executions. This possibility shows the
learning aspect of our prototype in addition to the improvement possibilities of our solution.

6. Related works

We find in literature other approaches applying software metrics to evaluate business process models based on the
similarities that exist between a software process and a business process5, 9.Based on these similarities, Elvira Rolón
et al. in 2006 proposed9 a set of metrics to evaluate business process models by checking their complexity. This
proposal is based on the adaptation and application of the Framework “Framework for Modeling and Evaluation of
Software Processes”. In the same context, and also interested in the complexity perspective, Jorge Cardoso et al. in
2006adapted4 a set of quality metrics applicable in the field of software in order to propose a group of metrics for the
evaluation of process models. It became the reference for many other works carried out in the context of improving

Fig. 4. Recommendations obtained in the Execution space. Fig. 5.Details obtained in the Execution space (1).

the quality of business process models.


Moreover, the objective of the work7 carried out by Khlif et al. in 2009 is to propose a list of metrics inspired this
time from the metrics applicable on the object oriented software field to evaluate the quality of the models of BPMN.
This adaptation is based on the correspondences between BPMN and object oriented software core concepts. The
contribution of the work in3 aims to propose an approach to help designers to improve the quality of their process
models (not only the syntactic quality but also the semantic and the pragmatic quality) by exploiting domain
knowledge. The work presented in6 deals with the management of the evolution of business process. This thesis
proposes an approach for checking the good progress of the business process after every change by evaluating the
impact of this change. The work realized in1presents an approach for the verification of business process models from
event logs, based on the process mining method. The goal of the work presented in8 is to provide a better understanding
in the field of process models complexity and to provide an overview of the corresponding metrics. In 10, authors
identified relevant indicators and threshold values, which are able to distinguish different levels of process model
quality.
850 Ghada Haouari et al. / Procedia Computer Science 121 (2017) 843–851
Ghada Haouari and Sonia Ayachi Ghannouchi / Procedia Computer Science00 (2017) 000–000

This overview of related work confirms, on the one hand, that improving the quality of business process models
has been a very active topic. On the other hand, it allows us to note that none of these approaches uses dynamic
metrics. This is the objective of our research.

7. Conclusion and future work

In this paper, we have focused on evaluating the quality of an emergency care process, more precisely the
corresponding BPMN model. Therefore, our work aims to propose a guiding approach for improving the quality of
this model by first evaluating the quality of the emergency care process BPMN model at the modeling phase as well
as at the execution phase.

Second, this guiding approach generates a set of recommendations for designers to improve the quality of this
BPMN model.
For this purpose, we adopted an extension of the BPM lifecycle. This extension demonstrates the improvements
brought by our solution. On the one hand, it is based on a set of quality metrics (static and dynamic) to evaluate the
model. On the other hand, it refers to a set of recommendations that can help to solve the defects and thus to improve
the model.
Our future work focuses on:
 Applying to our emergency care process other quality metrics.
 Reaching a version with high quality for our model of emergency care process. This version can then be
considered as a reference model and be reused by other institutions.
 Applying “Data Mining” techniques to determine threshold values for static metrics. In this paper, the threshold
values are determined by experts of modeling.

References

Fig. 6. Details obtained in the Execution space (2). 1.Allani O, and Ayachi S.
Verification of BPMN 2.0
Process Models: an Event Log-based Approach.Procedia Computer Science 100 (CENTERIS), Porto Portugal, October 5-7 2016. p.1064-1070.
2. AyachiGhannouchi S, and Ghannouchi S. Une expérience de BPR dans un hôpital tunisien. Systèmes d'information & management 2008;13:1:
89-116.
3. Ayad S. Business ProcessModelsQuality: evaluation and Improvement. Ecole Doctorale Informatique, Télécommunications et Electronique,
Paris I, Doctoral Thesis ; 2013.
4. Cardoso J, Mendling J, Neumann G, and Reijers H.A. A Discourse on Complexity of Process Models (Survey Paper). International Conference
on Business Process Management,Springer, Berlin Heidelberg; 2006. p.117-128.
Ghada Haouari et al. / Procedia Computer Science 121 (2017) 843–851 851
Ghada Haouari and Sonia Ayachi Ghannouchi / Procedia Computer Science00 (2017) 000–000

5. Cardoso J, Vanderfeesten I, and Reijers H.A. Computing coupling for business process models, 2010. available from Internet< http://eden. dei.
uc. pt/~ jcardoso/Research/Papers/Old% 20paper% 20format/Caise-19th-Coupling-Cardoso-Vanderfeesten. pdf>
6. Kherbouche M.O.Contributions à la gestion de l’évolution des processus métiers. Littoral Côte d’Opale University, Doctoral Thesis;2013.
7. Khlif W, Makni L, Zaaboub N, and Ben-Abdallah H. Quality metrics for business process modelling.Proceedings of the 9th WSEAS international
conference on Applied computer science, World Scientific and Engineering Academy and Society (WSEAS);2009. p.195-200.
8. Polančič G, and Cegnar B. Complexity metrics for process models–A systematic literature review. Computer Standards &
Interfaces2017; 51:104-117.
9. Rolón E, Ruiz F, García F, and Piattini M. Applying software metrics to evaluate business process models. CLEI-Electronic Journal2006 ;9:1.
10. Sanchez-González L, García F., Ruiz , and Piattini M. A case study about the improvement of business process models driven by indicators.
Software and Systems Modeling 2017, 16: 3: 759–788
11. Sommerville I.Software Engineering, 9th edition, Addison-Wesley, 2011.

You might also like