Evaluating the Process Capability Ratio of Patients’ Pathways by the

Application of Process Mining, SPC and RTLS

Sina Namaki Araghi, Franck Fontanili, Elyes Lamine, Nicolas Salatge, Julien Lesbegueries,

Sebastien Rebiere Pouyade and Frederick Benaben

Industrial Engineering Center of IMT Mines Albi, University of Toulouse, Albi 81000, France

Keywords:

Real-Time Location Systems, Process Mining, Statistical Process Control, Patient’s Pathways.

Abstract:

Learning how patients receive their health treatments is a critical mission for hospitals. To fulﬁll this task,

this paper deﬁnes patients’ pathways as business process models and tries to apply process mining, real-

time location systems(RTLS), and statistical process control (SPC) as a set of techniques to monitor patients’

pathways. This approach has been evaluated by a case study in a hospital living lab. These techniques analyze

patients’ pathways from two different perspectives: (1)control- ﬂow and (2)performance perspectives. In order

to do so, we gathered the location data from movements of patients and used a proof of concept framework

known as R.IO-DIAG to discover the processes. To elevate the performance analyses, this paper introduces

the process capability ratio of the patients’ pathways by measuring the walking distance. The results lead to

the evaluation of the quality of business processes.

1 INTRODUCTION

Healthcare organizations are facing the challenge of

improving the quality of their services continuously

(Rebuge and Ferreira, 2012). Consequently, these or-

ganizations need more knowledge about the executi-

ons of their processes. In the context of digital health,

this paper emphasizes the possibility of enriching the

knowledge of healthcare experts regarding to the real-

time state of patients by analyzing their location data.

The movements of patients inside healthcare orga-

nizations as they are executing different tasks, could

be treated as business process models and deﬁned as

patients’ pathways. To explain this statement, we in-

dicate that the authors in (Vanhaecht et al., 2010) de-

ﬁned the term of care pathway as a complex inter-

vention for the mutual decision-making and organi-

zation of care processes for a well-deﬁned group of

patients during a well-deﬁned period. It has also been

signiﬁed that the aim of these care pathways is to en-

hance the quality of care across the continuum by im-

proving risk-adjusted patients’ outcomes, promoting

patient’s safety, increasing patient’s satisfaction, and

optimizing the use of resources. Based on these deﬁ-

nitions and applied analyzing techniques in this paper,

it could be inferred that patients’ pathways can be ca-

tegorized as care pathways too.

On the other hand, each patient’s pathway con-

sists of the sequence of events, several steps, deci-

sion points, actors, and activities with the objective of

delivering health care to the patients. Consequently,

one could conceive these pathways as business pro-

cess models. Similarly, Dumas et al mentioned in

(Dumas et al., 2013) , a business process could be

seen as a mean that organizations use to deliver a ser-

vice or product to clients and it is constructed from

several decision points, sequences of activities, and

actors’ interactions.

As a service organization, a hospital needs to en-

sure about the quality of its services. Primarily, the

quality of hospitals’ services are evaluated by the du-

ration of their processes and activities. In this paper,

we evoke the idea of using the distance of patients’

pathways as a variable for assessing the quality of ser-

vices. There are several motives behind this choice.

For instance, there are several risks regarding transfer-

ring a critically ill patient. These distances should be

monitored precisely in order to be either minimized

or stabilized. Needless to mention that the duration of

processes could be correlated to the distance a patient

should take inside the facility. Additionally, thanks

to the proposed methods here, hospitals can acquire a

target value for monitoring the distance of patients’

pathways. This could lead to the detection of varia-

tions within patients’ processes and consequently en-

hancing the quality of processes.

302

Namaki Araghi, S., Fontanili, F., Lamine, E., Salatge, N., Lesbegueries, J., Rebiere Pouyade, S. and Benaben, F.

Evaluating the Process Capability Ratio of Patients’ Pathways by the Application of Process Mining, SPC and RTLS.

DOI: 10.5220/0007380803020309

In Proceedings of the 12th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2019), pages 302-309

ISBN: 978-989-758-353-7

This paper uses DIAG approach in order to extract

comprehensive knowledge from patients’ pathways

within hospitals. This approach which has been

shown in ﬁgure 1 is an updated version of the pre-

vious one in (Araghi et al., 2018). It consists of seven

main functions, that could be executed through four

different states. These states are Data state, Infor-

mation state, Awareness, and Governance. The se-

ven functions in this approach would accomplish a

goal which is transforming the location data of the

patients into decisions. This approach is being de-

veloped within R.IO-DIAG application. Each state

of the approach encompasses one or several functi-

ons. The ﬁrst function is the conﬁguring the envi-

ronment and the systems. This function is concerned

with installation of location systems and importing

the primary information into the R.IO-DIAG appli-

cation. These information are: identifying zones of

the experiment, patient’s information, and functions

that could be executed in certain zones. The second

function is the location data gathering. After obtai-

ning the primary event logs (cf ﬁgure 2(a)) a series of

interpreting rules have been devised in order to extract

the corresponding activities at each zone of a hospital.

A more detailed illustration of this function is inclu-

ded within (Namaki Araghi, 2018) article. After re-

ceiving the interpreted data, the list of un-linked acti-

vities would go within the process modeling function.

Thanks to several process mining algorithms, a car-

tography of process models will be discovered and

presented by declarative modeling languages or the

OPC modeling language. These models and its me-

trics (time and distance) will be studied by the process

analyzing function. After detecting the variations of

process models, a series of cause and effect diagnoses

will be performed in the process diagnosing to extract

the cause of those variations. Finally, at prognosing

function, thanks to discrete event simulation (DES)

several scenarios will be generated for choosing the

best scenario to improve the processes.

The next section provides an overview of process

mining, RTLS and the related works. However, the

third section presents the main focus of this paper

which is on the analyzing function of DIAG appraoch,

where we aim at evaluating the capability of processes

in order to ensure the quality of services. The fourth

section evaluates and illustrates the research work by

an experiment. Finally, the last section concludes the

paper.

2 BACKGROUND

There are two problematics that motivate experts to

consider patients’ pathways as business processes.

These problematics are social and technical. Con-

cerning the social aspect, Guardian issued a study in

UK (Pinchin, 2015) based on data from 2013 that

approximately 7 million hospital appointments have

been missed due to patients being lost in the facilities

and each costs on average £108. The impact of faci-

lity design on patients’ safety is a serious issue and

is related to the distance that patients walk in the fa-

cility. Therefore, monitoring the distance traveled by

patients helps hospitals to reduce these errors. Alt-

hough, measuring the exact value is only attainable

by using proper techniques and technologies such as

RTLS.

In the context of smart healthcare, the application

of RTLS is growing rapidly as a prerequisite. These

systems consist of three main components that com-

municate with each other thanks to radio-frequency

signals. The ﬁrst component is a tag which could be

attached to each object that need to be located. An-

tennas are the second component. They ﬁnd the po-

sition of the tags over the location area. The third

component is a location engine. This software uses

different types of algorithms and localization techni-

ques to calculate the positions of tags. Some of these

techniques are: triangulation, trilateration, Angle of

Arrival (AOA), Time Difference of Arrival (TDOA),

and Received Signal Strength Indicator (RSSI) (Co-

tera et al., 2016).

2.1 Process Mining

Recently, the application of process mining has been

developed in the context of healthcare (Rojas et al.,

2016) as an evidence-based methods to provide a fast,

automatic and efﬁcient way to map the processes.

The concentration of process mining is on discove-

ring knowledge from event logs which are registered

in an information system. Nowadays, the knowledge

generated by process mining is based on three acti-

vities: process discovery, conformance checking and

enhancement of business processes (Aalst, 2016). Se-

veral researchers have deﬁned process mining as a

practice which is being derived from the ﬁeld of data

mining (Tiwari et al., 2008). However, Van der Aalst

deﬁnes process mining as a bridge between process

science and data science (Aalst, 2016). He identi-

ﬁes that process mining and data mining both start

from data, however, data mining techniques are not

typically “process-centric”. Concerning process mi-

ning in healthcare, it mainly received attentions to-

Evaluating the Process Capability Ratio of Patients’ Pathways by the Application of Process Mining, SPC and RTLS

303

Configuring

the

environment

& systems

Location data

gathering

Data

Interpreting

Prcoess

Modeling

Process

Analyzing

Process

Diagnosing

Process

Prognosing

Layout map

Environment

characteristics

Zone

functionalities

Functions

characteristics

Actors

characteristics

R.IO-DIAG

Localization

systems

Primary

Event-logs

Localization

Systems

Interpreted events

corresponding to

the activities

Patients’

pathways

R.IO-DIAG

Variations

Performance

evaluation

(Quantititative

analyses)

R.IO-DIAG

Detected

causes

R.IO-DIAG

Scenarios

Data State

Awarensess

Information

State

Governance

Location

data

interpreting

rules

Objects'

movements

Individual

OPC’s

Time and

distance

Decision

Environment’s

Identified

and Categorized

(existing) information

Figure 1: DIAG approach for extracting knowledge from location event logs.

wards designing and applying discovering algorithms.

The enhancement and performance analysis are part

of process mining techniques that have not evolved

signiﬁcantly mainly in the area of patients’ pathways

(Partington et al., 2015). The focus of this paper is

on this gap. Therefore, the question for this research

work is: how hospitals can measure the quality of pa-

tients’ pathways?

Most of the applied methodologies in this area

contain three main activities: (i)cleaning and prepa-

ring the data, (ii) using discovery algorithms and (iii)

analyzing strategies (Rojas et al., 2016). As exam-

ples of these works, the authors in (Partington et al.,

2015) performed a research in Australia to evaluate

the performance of a care pathway in different hospi-

tals. They analyzed the processes based on the num-

ber of patients who have been admitted, the throug-

hput time of activities and the frequency of used he-

althcare procedures. They also provided an evaluation

of process mining literature in healthcare. Authors

highlighted a large gap among process mining activi-

ties. Based on their evaluation, only 3.5 % of research

works are concerned with performance evaluation and

enhancement of processes. They indicated that 82 %

of works are oriented towards process discovery and

22% are related to conformance checking. The asso-

ciation of process mining and RTLS did not receive a

lot of attention. However, it showed promising results

in previous works. In this regard, the current appro-

aches are highly dependent on the type of the con-

ﬁgured location system. In (Fernandez-Llatas et al.,

2015), Fernandez-Llatas et al. deﬁned seven steps to

transform location data in order to reach an enhan-

cement phase of processes. These steps have been

deﬁned in the context of a tool which transforms the

location data into the model-based analysis. They il-

lustrated their research by the PALIA-ILS SUITE ap-

plication (http://pmuc.ing.puc.cl/). In (Yang and Su,

2014), Yang and Su, reviewed the same mentioned

works in (Fernandez-Llatas et al., 2015) and empha-

sized the need to improve the performance evaluation

aspect of their work.

Most of these research works focused on the

control-ﬂow perspective. The main interest was to

map the patients’ pathways through process disco-

very. However, this paper expands this area by adding

quantitative analysis. This would help experts to not

only have a vision of the processes’ executions, but

also to acquire a real-time performance perspective of

the patients’ pathways by controlling the variations

within processes.

HEALTHINF 2019 - 12th International Conference on Health Informatics

304

3 METHOD

To ﬁll the mentioned gaps, this work tends to ex-

tend the DIAG approach presented in (Araghi et al.,

2018) by adding control charts and process capabi-

lity analysis to a process mining application in order

to analyze the real-time location data and quality of

patients’ pathways. This approach supports the per-

formance and control-ﬂow perspectives of a process

mining application in healthcare.

3.1 Statistical Process Control(SPC)

SPC is a powerful collection of problem-solving tools

useful in achieving process stability and improving

capability through the reduction of variations in the

process. SPC has already made its way into the healt-

hcare sector (Thor et al., 2007). However, it has been

used mainly for analyzing biological experiments but

not in a sense of analyzing patients’ pathways. The

Shewhart control charts are one of the most sophi-

sticated techniques of SPC. A typical control chart

has three indicators which are known as center-line

(CL), upper control limit (UCL), and lower control li-

mit (LCL). These lines are being represented in the

charts as horizontal lines. The UCL and LCL are in-

dicating that if a process is in control, then nearly all

of the sample points would fall between them. As

long as all of the points of the samples are between

LCL and UCL, no action is necessary. But, if a point

falls beyond those limits, it could be inferred that the

process is out of control due to the high level of va-

riations. Therefore, some inspections on different as-

pects of the process are required. There are several

types of control charts, such as ¯x chart, R-chart (range

chart), P-chart, and C-chart. The application of each

of these charts depends on the types of data and ana-

lysis that one could require. In this research work, ¯x

and R charts will be used since the outcomes are ba-

sed on two types of numeric data: Time and Distance

of patients pathways. In followings, the mathematical

principles for constructing the control limits will be

presented.

Let ¯x

, ¯x

,.., ¯x

, be the average values of the sam-

ples, then the process average is the grand average (

¯x):

¯x =

¯x

+ ¯x

+ .. + ¯x

(1)

If the range of each sample equals to R then:

R = x

max

− x

min

(2)

Now letR

, R

, ..., R

be the ranges of samples then the

average range of process is:

R =

+ R

+ ... + R

(3)

The control limits to construct the ¯x and R-charts are

as follows:

UCL

¯x

¯x + A

R UCL

¯x

¯x CL

R (4)

LCL

¯x

¯x − A

R LCL

The constant A

, D

, and D

in (4) changes based

on the size of the samples. These constants exist in

most of the mathematical statistics references (Jogle-

kar, 2005).

3.1.1 Process Capability Ratio

Another method to analyze the performance of pro-

cesses is in terms of process capability ratio (PCR)

or C

. PCR is a statistical metric for making a com-

parison between the output of a process and with the

speciﬁcations limits of the process. A process which

all of its outcomes fall between the speciﬁcation li-

mits is considered as a capable process. For exam-

ple, if hospitals deﬁne certain speciﬁcations as the ex-

pected length of stay for patients’ pathways; C

ratio

helps them to evaluate their performance. PCR ana-

lysis could be deﬁned by adjusting two new limits as

Upper Speciﬁcation Limit (USL) and Lower Speci-

ﬁcation Limit (LSL). These limits are speciﬁcations

relevant to the quality characteristics that one desires

to analyze (such as reliability of a process). In this re-

search work, the USL and LSL could be deﬁned ma-

nually by the healthcare experts, or be calculated by

analyzing the distribution of the gathered data. Equa-

tion (5) shows the mathematical expressions to calcu-

late C

. Where σ is the standard deviation of samples.

Equation (6) shows how to calculate the speciﬁcation

limit regarding to the distribution of the data.

USL − LSL

6σ

(5)

USL =

¯x + 3σ

LSL =

¯x − 3σ (6)

could have three states, which help experts to ana-

lyze the capability of the As-Is processes: If C

< 1;

it means that process is using up more than 100 %

of the tolerance band which means the process is not

capable to provide the desired outcome continuously.

If C

= 1; it means that process is using 100% of its

tolerance band. This implies that process may pro-

vide some undesirable outcomes, but statistically is

predictable and capable of satisfying the current spe-

ciﬁcation deﬁned by the organization. If C

> 1; the

process is using much less than 100 % of its tolerance

band. As a result, relatively few undesirable outcomes

could be produced by the process. These analyses can

be seen concretely by the description of the study case

in the next section.

Evaluating the Process Capability Ratio of Patients’ Pathways by the Application of Process Mining, SPC and RTLS

305

4 CASE STUDY AND THE

EXTRACTED RESULTS

To validate this approach, an experiment in a hos-

pital living lab has been conducted. In this regard,

R.IO-DIAG application has been developed within

R.IO SUITE platform (https://research-gi.mines-

albi.fr/display/RIOSUITE/R-IOSuite+Home) as a

proof of concept, and for the illustration of the

results. This tool receives the location data, reﬁnes

them and generates business process models with the

addition of robust statistical analyses and diagnoses.

The results of this study case are the outcomes of two

core functions of DIAG approach: process modeling

and analyzing. This experiment had been carried

out during ten days, and 150 patients have been

monitored. They have been divided into 10 samples

with the unique size of 15. These patients had similar

proﬁles regarding which health care procedure they

required. Event logs from the RTLS have been

generated by the location engine with the JSON

format. Each event contains several information

such as event ID, time-stamp of entering a zone or

exiting one, location data and other complementary

data such as room temperature, humidity and tag’s

battery level. Figure 2 shows an example of patients

tracking view and the primary event logs. To perform

process discovery, a set of complex event processing

rules have been deﬁned based on DIAG reference

model (Namaki Araghi, 2018). This reference model

helps to identify which types of activities could

happen in a certain zone. The details of these rules is

within location data interpreting function of DIAG,

which are evoked in (Namaki Araghi, 2018). After

gathering the location data of pathways, the modeling

function is the next step to gain a view on the way

processes are being executed.

4.1 A Control-ﬂow Perspective

The ﬁrst step to obtain a model-based analysis of pa-

tients’ pathways is to perform automatic process dis-

covery. There are several existing process discovery

algorithms (Augusto et al., 2017). DIAG uses sta-

ble heuristic mining algorithm since it can be app-

lied for a collective modeling of pathways (describing

the algorithm in detail is beyond the limits of this

article). As it has been shown in ﬁgure 3 a qualita-

tive analysis can be extracted by illustrating patients’

pathways as process models. After analyzing the mo-

del, it has been indicated that most of the cases in the

model have ignored one important step in their pro-

cesses. This step was the "UROLOGY_checkout".

The primary diagnosis indicated this could be due

to the long waiting period at the "waiting area for

checkout_UROLOGY". Such qualitative analysis can

be more useful for the organizations if it is being sup-

ported by quantitative analyses too. Existing mathe-

matical analyses within process mining tools do not

evaluate the quality of processes. The performance

perspective section of this paper targets this proble-

matic.

4.2 The Performance Perspective

It is obvious that the pathways could differ for each

patient. However, it is a non-trivial task for hospi-

tals to offer a stable service quality to patients. As

Montgomery emphasizes in (Montgomery, 2007), the

quality of services which patients are conceiving is

directly inﬂuenced by the stability of processes. Du-

ring this experiment, we encountered that patients did

not have problems with the waiting times or length of

pathways in particular but they expressed their dis-

satisfaction when they had to spend longer periods

of time than expected in order to perform certain

tasks. Additionally, they had difﬁculties for ﬁnding

the rooms in the hospital. Therefore, we proposed to

use process control charts to enhance the performance

by discovering and diagnosing the variations.

Since the control limits on the ¯x-chart (Figure 4b)

depend on the process variability, it is ideal to begin

with the R-chart(Figure 4a). Unless the process vari-

ability is in control, these limits (in ¯x-chart) will not

have much meaning (noa, ) and the process is already

out-of-control.

The R-chart helps to ensure the stability of the ex-

tracted data. The CL is the average of all the sub-

groups’ ranges. The other control limits are set by a

distance of 3σ (standard deviation) above and below

the center line. These thresholds deﬁne the limits for

expected variations in the subgroups ranges. The ﬁ-

gure 4a shows R-chart for analyzing the range of dis-

tances for patients’ pathways. Based on the stability

in ranges, we may now construct the ¯x-chart presen-

ted in ﬁgure 4b. This ﬁgure presents the instability

of the average traveled distance by patients. The red

points in the ¯x-chart indicate that within three days

of the experiment, there were some assignable causes

that affect the distance of pathways for the patients.

These variations in the process have been investiga-

ted by the help of healthcare professionals. Because

of an increase in the number of admitted patients on

those days, the department had the lack of resources

to perform medical examination for all the patients.

Therefore, nurses asked some patients to report their

problems to other exam rooms in the hospital. This

caused the traveled distance of some pathways exceed

HEALTHINF 2019 - 12th International Conference on Health Informatics

306

(a) An example of the primary event log. (b) A view on the movement of the patients.

Figure 2: Primary results of tracking patients by RTLS.

Figure 3: The extracted process model by R.IO-DIAG.

from the norm value. Consequently, this issue led to

an instability in the process.

The length of patient’s pathway is a practical indi-

cator for the capability of a hospital in providing efﬁ-

cient services. The reason could be seen as the effect

on the length of stay in the hospital. Also, it inﬂuen-

ces the efﬁciency of providing emergency treatments

to the exact location of a patient with a critical sta-

tus. Thus, this research work seeks different means to

analyze the capability of processes based on the deﬁ-

ned CTQ (Critical To Quality) speciﬁcations. These

speciﬁcations (USL, target, LSL) either are deﬁned

by the health professionals or could be adapted from

the way patients are moving in the hospital. Figure 5

presents the results on process capability ratio (C

) of

the experiment.

There are several points that should be inferred

from the PCR analysis; ﬁrst of all, this chart consists

of two speciﬁcation limits that show the possible mar-

gins for the length of pathways. As it has been shown

in ﬁgure 5, there are processes that have the distance

higher than USL and lower than LSL. These proces-

ses are not satisfying the targeted speciﬁcations. The

hospital’s experts have deﬁned an USL of 220 meters

provisionally. The LSL has been adapted automati-

cally from the current state of processes with the de-

ﬁned formula in 6. The C

is less than 1 which shows

that processes are being executed in an inefﬁcient way

and they are not meeting the expectations. As it has

been shown, approximately 5 % of patients are wal-

king more than the upper speciﬁcation limit. Note that

l, and C

u, demonstrate the performance of proces-

ses near to the lower speciﬁcation limit and the upper

speciﬁcation limit. C

k illustrates the capability of the

process when the average of the samples are not well-

centered. C

measure could be useful if we want to

use the average value as the target to reach. C

could

be applied to monitor the difference between the tar-

get value and the average of the results. For instance,

the target in this chart has been identiﬁed as 135.115

meters. However, the average length of pathways is

different (Center = 155.04). As process average mo-

ves off of the target, C

grows greater. This would

help to further analyses and diagnoses. The same ana-

lysis are applicable and have been done for the dura-

tion of processes which their explanations are beyond

the scope of this article. Relevant to the "enhance-

ment" activities of process mining paradigm, the pre-

sented methods can provide applicable means in order

to detect the process variations from the outcomes of

the processes.

5 CONCLUSIONS

This paper aimed at addressing several issues regar-

ding monitoring of patients’ processes in healthcare

organizations.

• Firstly, it emphasizes the need to consider pa-

tients’ pathways as business processes. Then

it provides technical and technological means to

monitor these pathways. The advantage of this

approach is related to a shorter period of data ex-

traction and the accuracy level of extracted data.

Evaluating the Process Capability Ratio of Patients’ Pathways by the Application of Process Mining, SPC and RTLS

307

Group

Group summary statistics

1 2 3 4 5 6 7 8 9 10

50 100 200

R Chart

for Distance_Of_Patients_Urology

LCL

UCL

Number of groups = 10

Center = 121.3

StdDev = 34.93664

LCL = 42.04047

UCL = 200.5595

Number beyond limits = 0

Number violating runs = 0

(a) Analyzing the variations of pathways’ length

Group

Group summary statistics

1 2 3 4 5 6 7 8 9 10

120 160 200 240

xbar Chart

for Distance_Of_Patients_Urology

LCL

UCL

Number of groups = 10

Center = 155.04

StdDev = 34.93664

LCL = 127.9782

UCL = 182.1018

Number beyond limits = 3

Number violating runs = 0

(b) The ¯x-chart to monitoring the stability of pa-

tients’ pathways

Figure 4: Control charts analysis related to the distance of pathways.

100 200 300 400

Process Capability Analysis

for Distance_Of_Patients_Urology

LSL USLTarget

Number of obs = 150

Center = 155.04

StdDev = 34.93664

Target = 135.115

LSL = 50.23008

USL = 220

Cp = 0.81

Cp_l = 1

Cp_u = 0.62

Cp_k = 0.62

Cpm = 0.704

Exp<LSL 0.13%

Exp>USL 3.1%

Obs<LSL 0%

Obs>USL 5.3%

Figure 5: Evaluation of the process capability by the length

of patients’ pathways.

• Moreover, the presented approach provides a new

sight to evaluate the capability of patients pro-

cesses based on the traveled distance of their

pathways. To the best of our knowledge regarding

the literature, analyzing the stability and capabi-

lity of patients’ processes by the traveled distance

of their pathways have never been done.

• Most of process mining research works are focu-

sing on a challenge of how to discover a proper

and ﬁtting process model, which is an inevitable

task. However, in this paper we tried to give an

extra attention to the enhancement phase of pro-

cess mining which we believe is necessary in or-

der for process mining to become more applicable

for healthcare organizations.

In this paper care-ﬂow and performance perspectives

are provided thanks to the framework of R.IO-DIAG

which is exclusively being used to extract business

process models from location data.

As it has been introduced, R.IO-DIAG applies dif-

ferent statistical process monitoring techniques which

are not being used in other process mining packages.

Thanks to these techniques, it is possible to moni-

tor and reduce the variability of patients’ pathways,

which could lead to the quality improvement of pro-

cesses.

After analyzing the patients’ pathways and high-

lighting the shortcomings of the processes, the future

perspective of this research work is to provide an au-

tomatic diagnosis of the processes.

REFERENCES

Aalst, W. M. P. v. d. (2016). Process Mining: Data

Science in Action. Springer. Google-Books-ID: hU-

EGDAAAQBAJ.

Araghi, S. N., Fontanili, F., Lamine, E., Tancerel, L.,

and Benaben, F. (2018). Applying Process Mining

and RTLS for Modeling, and Analyzing Patients’

Pathways. pages 540–547.

Augusto, A., Conforti, R., Dumas, M., La Rosa, M., Maggi,

F. M., Marrella, A., Mecella, M., and Soo, A. (2017).

Automated Discovery of Process Models from Event

Logs: Review and Benchmark. arXiv:1705.02288

[cs]. arXiv: 1705.02288.

Cotera, P., Velazquez, M., Cruz, D., Medina, L., and

Bandala, M. (2016). Indoor Robot Positioning Using

an Enhanced Trilateration Algorithm. International

Journal of Advanced Robotic Systems, 13(3):110.

Dumas, M., Rosa, M. L., Mendling, J., and Reijers, H. A.

(2013). Introduction to Business Process Manage-

ment. In Fundamentals of Business Process Mana-

gement, pages 1–31. Springer, Berlin, Heidelberg.

Fernandez-Llatas, C., Lizondo, A., Monton, E., Benedi, J.-

M., and Traver, V. (2015). Process Mining Metho-

dology for Health Process Tracking Using Real-Time

Indoor Location Systems. Sensors, 15(12):29821–

29840. WOS:000367539100004.

Joglekar, A. M. (2005). Appendix G: Table of Control Chart

Constants. In Statistical Methods for Six Sigma: In

R&D and Manufacturing, pages 311–311. John Wiley

& Sons, Inc.

Montgomery, D. C. (2007). Introduction to statistical qua-

lity control. John Wiley & Sons.

HEALTHINF 2019 - 12th International Conference on Health Informatics

308

Namaki Araghi, S. (2018). ACS/IEEE International

Conference on Computer Systems and Applications

AICCSA.

Partington, A., Wynn, M. T., Suriadi, S., Ouyang, C., and

Karnon, J. (2015). Process mining for clinical proces-

ses: A comparative analysis of four Australian hospi-

tals. ACM Transactions on Management Information

Systems, 5(4):19:1–19:18.

Pinchin, J. (2015). Getting lost in hospitals costs the NHS

and patients. The Guardian.

Rebuge, l. and Ferreira, D. R. (2012). Business pro-

cess analysis in healthcare environments: A methodo-

logy based on process mining. Information Systems,

37(2):99–116.

Rojas, E., Munoz-Gama, J., Sepúlveda, M., and Capurro,

D. (2016). Process mining in healthcare: A literature

review. Journal of Biomedical Informatics, 61:224–

236.

Thor, J., Lundberg, J., Ask, J., Olsson, J., Carli, C., Hären-

stam, K. P., and Brommels, M. (2007). Application of

statistical process control in healthcare improvement:

systematic review. Quality & Safety in Health Care,

16(5):387–399.

Tiwari, A., Turner, C. J., and Majeed, B. (2008). A re-

view of business process mining: state-of-the-art and

future trends. Business Process Management Journal,

14(1):5–22.

Vanhaecht, K., Sermeus, W., Peers, J., Lodewijckx, C., De-

neckere, S., Leigheb, F., Decramer, M., and Panella,

M. (2010). The impact of care pathways for exacer-

bation of Chronic Obstructive Pulmonary Disease: ra-

tionale and design of a cluster randomized controlled

trial. Trials, 11:111.

Yang, W. and Su, Q. (2014). Process mining for clinical

pathway: Literature review and future directions. In

2014 11th International Conference on Service Sys-

tems and Service Management (ICSSSM), pages 1–5.

Evaluating the Process Capability Ratio of Patients’ Pathways by the Application of Process Mining, SPC and RTLS

309