U.P.B. Sci. Bull., Series D, Vol. , Iss. , 20 19 ISSN 1223 -7027 [629306]

U.P.B. Sci. Bull., Series D, Vol. …, Iss. …, 20 19 ISSN 1223 -7027
IMPROVING HEALTHCARE INDUSTRY USING MINING
TECHNIQUES

MAHA Z AYOUD1, SORIN IONESCU2
ABSTRACT : Healthcar e is an important sector nowadays; it is related to many entities which
needs the professionals to meet the different requiremen ts. However, there are many issues
appears in healthcare domain that attract the researchers to use different methods and techn iques
such as process mining, which is the practice of extracting knowledge in a specific organization
and provides an accurate image of any system and suggests solutions to develop it. This paper
defines the healthcare in general with its main needs , defines the main process mining techniques ,
highlights the β algorithm and its impact on improving the healthcare from many perspectives .

Keywords : Healthcare, Business Process, Economics .
1. Introduction
Industry is the neural system in any country, and the growth of the industrial domains make the
technologies, and the methods that can handle this growth is mandatory. However, the industrial
domains are not limited to companies and organizations only, but they mean also the se rvice
sectors as well, such as the healthcare service sector. Healthcare is an important field of study
because if its impact on the whole society and politics. Healthcare includes many aspects to study,
and it is an attractive domain for researchers to fi nd new methods of continues improvements .
The process mining is an interesting field of study for both researchers and practitioners, and many
algorithms and methods are invented due to the importance of this field of study [1] , but many
methods are still facing problems when they are implemented in real life applications due to the
complexity of some systems and their information management . There are many challenges and
issues can be resolved by using one of the invented process mining techniques , such a s the
challenge of cost management in any domain, customer satisfactions, staff satisfactions, and
finally the continues performance improvements. Using any of the process mining algorithms or
methods requires h aving management information system that can store data of th at organization ,
hence the process mini ng methods can discover the existed process and study its limitations and
strengths . As a challenge of using the process mining methods is choosing the right method among
all the existed methods to di scover knowledge that can provide the organization with the required
steps toward improvement [2]. However, the process mining techniques aim to develop, and
discover the process based on the information extracted from event logs [4]. It is a result of the
need of emerging between the data mining [ 3] and business process management [ 3], where data
mining concerns with analyzing the different types of the data sets and the business process
management focuses on modeling these data types and files. Hence, the process mining plays the
middleware role between the two, to combine the data analysis with modeling. Moreover, process
mining can handle the raw data or the event logs of any organization. There is no accepted

1 College of Engineering and Technology , American University of the Middle East. Kuwait , email: maha .zayoud @
aum.edu.kw
2 Industrial Engineering Department, Polytehnica University of Bucharest, Romania , e-mail: sorin. ionescu @upb.ro

Maha Zayoud , Sorin Ionescu

benchmark to evaluate and compare the di fferent proposed process mining algorithms which
makes it difficult to select a suitable process mining algorithm for a given enterprise or application
domain [4]. Process mining algorithms are used to mine business process models using process
logs. The min ed models are compared against the current process models of the enterprise for
conformance checking that is a basic step of process mining. The n ext step , is to discover more
efficient, streamlined business process models [2].

2. Healthcare System
The structure of health care system consists of medical care services under three different parties:
primary care, hospitals, and community healthcare [ 5]. Each type aims to provide a specific kind
of medical care, and each one is offered by the specialists’ doctors , or any other staff who is
responsible for this type of care service. The goal of the healthcare system is that good healthcare
should be available to all people regardless of their financial situation. Providing preventive health
services is the major function of the community health service [5] . Funding the healthcare sectors
is one of the challenges that helps to ensure the quality services. The community health service
refers to the cooperation with the local government in order to enable health and personal public
care [6] . However, t he healthcare system is funded from the general taxation system and the
national insurance contributions [6]. These funds are div ided among the health authorities to cover
the medical expenses of each area. Although , the healthcare system is improved successful ly, but it
suffers from several problems such as shortage of resources, very demanding hospitals, waiting
lists for low acui ty patients and bad management. In the other hand, the advantages are worth to be
highlighted , such as following the international standards for healthcare , and the public services
are non-expensive service which grasp the attention of majority of individu als. Resources are
directed by the system toward different areas and several types of care based on the region needs.
In parallel, primary care and hospital treatment is also covered by the private medical sector. Most
private care are part of private medi cal insurance program. Healthcare is not an easy domain to
study or develop because of the wide range of entities that consists of, or the different services that
takes care of them, such as the medications parts, service cost, staff satisfactions, resources,
researches, and dat a. As an example of one part of healthcare system is, hospital which is one
major and top-level healthcare system, and can be divided into different departments. These
departments are interacting with each other and sharing a variety of professional resour ces
(Doctors, Nurses, Staff, etc.). This crossover leads to a very complex and sensitive system as
mentioned because of the different departments and units in a hospital [ 6].

3. Process Mining Techniques

3.1. Business Process:

As seen previously that process mining is representing the middle step between the data mining,
and the business process management. Data mining is defined a s the process of analyzing data
from different perspectives and summarizing it into useful information that can be us ed to i mprove
many economic factors. While, the b usiness process is defined as a collection of interrelated tasks,
performed to achieve a business outcome which is a chain of tasks from purchasing to
manufacturing to selling and delivering. Business proces s can be divided into two kind operational
which is related to the core business and management which include the information system and
strategic decisions [7]. Process innovation is the implementation of a new method for production
by developing new products or services or changing the business model, or by developing new

IMPROVING HEALTHCARE INDUSTRY USING MINING
TECHNIQUES
method of delivery. The innovation in any business can be achieved by changing the technology
and environment of any origination. The advantages behind the process inn ovation like improving
quality, faster processes, reduced labor costs, reduced materials, energy consumption and
environmental damage .

3.2. Process Mining Algorithms :
The process mining is the knowledge extracted from the event logs that is recorded by an
information system applied in the organization [3 ]. Although, the information systems help to get
the event logs, but these logs are rarely used to analyze the underlying processes. Hence, the
mining techniques aim to develop, and discover the process based on the information extracted
from event logs [ 3].It is a result of the need of emerging between the data mining and business
process management, where data mining concerns with large data sets and the business process
management focuses on modeling these dat a. Hence, the process mining plays an important role
between the two, to combine the data analysis with modeling. Moreover, process mining can
handle the raw data or the event logs of any organization. However , process mining algorithms are
mainly categorized as either local algorithms such as the α algorithm [8] and the heuristic [9], or
global algorithms such as genetic [10] and fuzzy algorithm [11].
Local algorithms are based on local information about event s in a log [ 12], which is used to detect
the dependencies between these events by extracting information about what tasks directly precede
or directly follow each other in that log. The global algorithms are defined as non -local constructs
that cannot be determined by only looking at the direct successors and predec essors which is the
local context of a task in a log [12].
The process mining algorithms and techniques face many challenges such as:
• Incomplete and noisy input data.
• Distinguishing sequences, fo rks, and concurrency.
• Deriving block -structured and arbitrary loops.
• Distinguishing repeated activities especially in the loops.
• Dealing with fuzzy process entry and end points.
• Detecting different process types and variants.
Managing a critical system suc h as the healthcare system requires a suitable algorithm which can
manage the above challenges to give an accurate process model. The well know algorithms of
process mining have many advantages and disadvantages, and many of them have extension like
the α algorithm that is simple and easy to use but it is not suitable for real life application, has an α
miner for example. Because of these challenges, the need to use a reliable algorithm is needed
especially in out domain which the healthcare system, therefo re the β algorithm is used in this
study and a brief explanation about it is mentioned in the next sections.
4. The β Algorithm
The β algorithm framework [ 12] is a process mining algorithm that is designed to overcome many
issues appear in other process mining methods. The β algorithm is a learning technique to extract
knowledge about any system by learning different log files of that system [12]. This framewo rk is
a probabilistic and it learns the process in an accumulative manner. The learning process is divided
into phases as shown in the flowchart of Figure 1. First, it starts with learning the set of events in

Maha Zayoud , Sorin Ionescu

the log files then, it calculates the dependen cies among those events. After learning the
dependencies, the probability distributions of those events and their correlation are detected [12].
A knowledge base is integrated to the learning framework to predict the coming events out of the
learned model. The knowledge base has few simple rules for real -time prediction of events. The β
algorithms starts with analyzing the data files first, where all required information about that
system is input to the framework, then the knowledge discovery starts, with the learning of the
events, dependencies, probabilities, and frequencies all will build the work flow model to reach at
the end what is called a predict model which describes the current process clearly. The β algorithm
is not a theoretical solution only; it is proved mathematically, which is the accepted method of
proving any new technique [ 12]. Then, it is written in an algorithmic format and it requires data in
terms of events (activities of all resources in the system), where each event should have a st arting
time and ending time to deploy the rules of this framework correctly [ 12].

Fig 1. The β algorithm Framework Flowchart [12 ].

5. Applying the β Algorithm Framework on Healthcare System

This section shows how the process mining methods is applied on real life application. The
example is data input represents a log file of patients’ activities in a hospital. The β algorithm is
chosen among other methods due to its many advantages that are presented in Figure 2.

IMPROVING HEALTHCARE INDUSTRY USING MINING
TECHNIQUES

Fig 2. Famous Process Mining Algorithms VS. Challenges [12 ].

Where
means this algorithm can handle the required challenge, and X means it cannot
handle it. As seen in Figure 2, the β algorithm can manage many challenges that face any
metho d to detect the process and all its aspects. This framework is chosen due to its advantages
in detecting the events in any system after learning the log file. Then, the β framework learns
how these events are related to other, such as which event follow an other event, and the different
logical re lations that may appear between events are also detected which helps to build the
knowledge base toward the existed process. The β framework is also a probabilistic approach as
shown in Figure 1, hence it suits the real-life applications more than other techniques to detect
high risk factors and the disadvantages of the current process [12]. The needs of using process
mining techniques, especially like the β algorithm which is suitable for healthcare system and
can b e applied on real data, because building a process from scratch is easier than trying to
detect the current process of an existed system and knowing all its aspects accurately, with all
advantages and disadvantages which open s the way for improvement and development of an
organization without causing huge loss of data or resources.

Fig 3. Events’ relations versus Process Mining Algorithms [ 12]

5.1. The Economic Benefits of Applying the β Framework in Healthcare

There are many advantages of using process mining methods , such as provid ing knowledge about
the existed process in the system, and how this process works based on evidence and using the log
data files that are stored in the management information system of an organization , which allows
an objective reconstruction of the process flows [ 13].

Maha Zayoud , Sorin Ionescu

The benefits of using Process mining algorithms are:

1. Time factor :
Discoveri ng the process activities need s several weeks or months, which is a considerable time in
any system, implement ing a process mining can come up with first results and hypotheses quickly,
which can help to increase trust and engage people with the services that are provided in that
hospital without consuming a considerable time using the traditional methods [13].

2. Understanding the system:

Process mining provides an objective reference on how things are done and giving the reason of
why people work the way they do, which is rarely detected using the observation or data analysis.
However, understanding the root causes of inefficiencies also on the human level is crucial to
successfully implement organizational change , hence it maximizes the value by getting deep
knowledge that is normally not discovered [ 1].

3. Get a head start within new domains
Process mining can help consultants, who are specialists in specific domain, and able to provide
assistant in their domain only, to understand the new domains also and approach their job by
having a tool to understand the process quickly , and increase the productivity especially of junior
analysts, right from the start [13].

4. Help Clients to justify Changes within the Company
Process mining can provide an objective reference for the clients of any organization , which can
put them in a stronger position to achieve their goals successfully [1 3], which getting the right care
in the case of the hospital.

5. Compare “before” and “after”
As mentioned before, the process mining can understand the process quickly, e specially for
process improvement projects , where we need to demonstrate the effect of the implemented
changes. For example, showing how the process has been streamlined after a change (and one
month of new data collection) by comparing the “before” and “after” images. Ideally, the process
performs much better n ow, and you can use process mining to communicate these results [1 3].
Hence, the above benefits show the impact of using the process mining techniques on any
organization especially by reducing the time of analysis, the cost of understanding the process, a nd
getting the objective references of clients’ goals and here are the patients and the medical staff as
well.
6. The β algorithm on hospital example
The β algorithm is implemented on a hospital that has some issues related to time consumption in
serving patients, cost, staff satisfactions and patients satisfactions as well. The following is the

IMPROVING HEALTHCARE INDUSTRY USING MINING
TECHNIQUES
case study of that hospital. The data input to the β platform is many log files that include one
patient’s activities with their timing information, and many patients’ events as well. Figure 4. And
Figure 5 Show examples about what are the basic events for a patient in that hospital.

Fig4. One Patient Log File [ 12]

Fig 5. Another Patient Log File [12]
The patient’s information that is categorized as events such as arriving to hospital, checking
urgent, billing, etc. all input to the platform to be processed and studied. Then the dependency
matrix between these events is an output result of this information such as Figure 6. Where the 1
means their dependency between this event located in the row and the other event located in the
column of the matrix, 0 means there is no dependency between these two events. As an example,
the billing event depends on discharging the patient after getting the requir ed treatment by the
doctor.

Maha Zayoud , Sorin Ionescu

Fig 6. Dependency Matrix of First Log File

Fig 7. Dependency Matrix of First Log File [ 12].
The β platform can know the relation between the events of patients, such as which event can start
with which event, and which event cannot start at the time the other one is executed . Moreover, the
frequency of the events and their relations is calculated using the platfor m, hence the probabilities
of these events are detected. In fact, the β algorithm [ 12] is an extension of the α algorithm [ 8],
which is originally builds a workflow nets out of the events log file and can use Petri nets [ 14].
Hence, the Petri net that is d iscovered from the output results of the β platform is presented in
Figure 8.

Fig 8. Petri Net for Patients Events in the Hospital.

IMPROVING HEALTHCARE INDUSTRY USING MINING
TECHNIQUES
After that, and from detecting all the required information and events and their relations and
probabilities, the proces s model is discovered as shown in Figure 9. This process model is a
starting point to understand the current system with all its advantages and disadvantages, such as
the deadlock in the system.

Fig 9. A Process Model for a Hospital [12].
7. Results Analysis of the Process Mining
Analysi ng the results means studying the strengths, weaknesses, opportunities and threats ,
identifying and analyzing the internal and external factors that can have an impact on the viability
of an organization , product, place or person [ 15]. Business entities are the main users of the
analysis, the analysis is used to show the benefits on the industry of using the process mining
techniques, since it shows the different factors that needs to be analyzed . Using a huge amount of
patient’s information that includes the patients activities and their starting and ending time with the
duration, shows the number of detected events and their probabilities which leads to know more
about the system and its events, h ence the deadlock can be detected and suggestion on better
solution can be implemented. Figure 10. Shows how many events are detected from a log file of
big number of patients.

Maha Zayoud , Sorin Ionescu

Fig 10. Events Statistics
Figure 11. helps to detect the probabilities of e ach event in the hospital system.

Fig 1 1. Probability of Events
8. Conclusion

As a summary, this paper aims to present the benefits of using the process mining techniques in
any organizations . It explains what the challenges are of using process minin g algorithms, in
addition it focusses on healthcare industry and presents its main issues. In this study the β
algorithm framework which is an extension of the well -known algorithm named the α algorithm is
implemented on hospital example because the hospital is a t op entity in the healthcare system.
Even though , there are many algorithms and methods in th e field of process mining but choosing
the right method for an organization is a challenge , hence the β algorithm is cho sen. The log files
information is input to the platform and the output results are simulated to build a knowledge base
about the system of the hospital. Some of the economic factors from using the process mining
techniques are mentioned also in this paper. Finally , the process mining is important to be
implemented in the domains that have issues or deal with big number of clients, this can lead to
improve their processes and increase the overall perfo rman ce and achieve the required goals of the
system. All the mentioned cannot be achieved without choosing the suitable method for an
organization .

IMPROVING HEALTHCARE INDUSTRY USING MINING
TECHNIQUES
R E F E R E N C E S
[1] Van Der Aalst, Wil. Process mining: discovery, conformance and enhancement of business
processes . Vol. 2. Heidelberg: Springer, 2011.
[2] Van Der Aalst, Wil, Arya Adriansyah, Ana Karla Alves De Medeiros, Franco Arcieri, Thomas
Baier, Tobias Blickle, Jagadeesh Chandra Bose et al. "Process mining manifesto." In International
Conference on Business Process Management , pp. 169 -194. Springer, Ber lin, Heidelberg, 2011.
[3] Laudon , Kenneth C., and Jane P. Laudon. Management information systems . Prentice Hall
PTR, 1999.
[4] Van Der Aalst, Wil. Process mining: discovery, conformance and enhancement of business
processes . Vol. 2. Heidelberg: Springer, 2011.
[5] Andersen, Ronald M., Thomas H. Rice, and Gerald F. Kominski. Changing the US health care
system: Key issues in health services policy and management . John Wiley & Sons, 2011.
[6] Chaulagai, Chet N., Christon M. Moyo, Jaap Ko ot, Humphrey BM Moyo, Thokozani C.
Sambakunsi, Ferdinand M. Khunga, and Patrick D. Naphini. "Design and implementation of a
health management information system in Malawi: issues, innovations and results." Health policy
and planning 20, no. 6 (2005): 375 -384.
[7] A.J.M.M. Weijters, W.M.P. van der Aalst, and A.K. Alves de Medeiros,” Process Mining with
the Heuristics Miner Algorithm” .
[8] Van der Aalst, Wil, Ton Weijters, and Laura Maruster . "Workflow mining: Discovering
process models from event logs." IEEE Transactions on Knowledge and Data Engineering 16, no.
9 (2004): 1128 -1142.
[9] Weijters, A. J. M. M., Wil MP van Der Aalst, and AK Alves De Medeiros. "Process mining
with the heuristics miner -algorithm." Technische Universiteit Eindhoven, Tech. Rep. WP 166
(2006): 1 -34.
[10] De Medeiros, AK Alves, and A. J. M. M. Weijters. "Genetic process mining." In Applications
and Theory of Petri Nets 2005, Volume 3536 of Lecture Notes in Computer Sc ience . 2005.
[11] Günther, Christian W., and Wil MP Van Der Aalst. "Fuzzy mining –adaptive process
simplification based on multi -perspective metrics." In International conference on business
process management , pp. 328 -343. Springer, Berlin, Heidelberg, 200 7.
[12] M. Zayoud, Y. Kotb, S. Ionescu, β Algorithm: A New Probabilistic Process Learning
Approach for Big Data In Healthcare. IEEE Access Journal PP (99):1 -1 June 2019 DOI:
10.1109/ACCESS.2019.2922635.
[13] https://fluxicon.com/blog/2010/05/7 -reasons -for-consultants -to-do-process -mining/ .
[14] Reisig, Wolfgang. Petri nets: an introduction. Vol. 4. Springer Science & Business Media,
2012.
[15] Sabbaghi, Asghar, and Ganesh Vaidyanathan. "SWOT analysis and theory of constraint in
information technology projects." Information systems education journal 2, no. 23 (2004): 3 -19.