Senior High School Department

Manila Campus

Developing a Smart Triaging System Application Based on Fuzzy Logic for

Patient Prioritization

Geronimo, Andrea Dominique G.

Leopoldo, Maria Mikaela P.
Ngo, Shaine Andrew C.
Tolentino, Gabriella Nicole F.

3rd Term, AY 2020-2021

<Date of Submission>

I
 Abstract

Hospitals in the Philippines have always dealt with issues of overcapacity in the emergency
department. One of the main problems is that a majority of emergency departments utilize a
manual triage system (paper based), which is prone to triage inaccuracy and increased workload.
The aim is to develop a smart triaging system that uses fuzzy logic for a faster and more accurate
method of triaging. The application is going to be made through the python programming
language and its basis on sorting patient severity is from the Emergency Severity Index (ESI).

Keywords: emergency rooms, triage systems, artificial intelligence

I
 Table of Contents

Abstract I

List of Tables IV
List of Figures V

1.0 Introduction 1
1.1. Background of the Study 1
1.2. Statement of the Problem 2
1.3. Research Objectives 2
1.4. Scope and Limitations 3
1.5. Significance of the Study 3

2.0 Review of Related Literature 4

2.1. Introduction 4
2.2. Emergency Rooms 4
2.3. Triage System 6
2.4. Artificial Intelligence 12

3.0 Theoretical and Conceptual Framework

17
3.1. Introduction 17
3.2. Conceptual Framework 17
3.3. Theoretical Considerations 18

19
4.0 Research Methodology
4.1. Materials 19
4.2. Procedure 19
4.3. Research Design 22
4.4. Data Analysis Strategy 27

5.0 Results and Discussions 29

5.1. Results and Analysis 1 XX

II
 5.2. Results and Analysis 2 XX
5.3. Results and Analysis 3 XX

6.0 Conclusion and Recommendations XX

References XX

Appendix
A. Gantt Chart XX
B. Estimated Budget XX
C. Informed Consent Form XX
D. Informed Assent Form (optional, depending on your participants) XX
E. Data Collection Instruments XX
F. Evaluation Instruments (optional, depending on your research design) XX
G. Transcript of Interview (optional, depending on your research design) XX
H. Evaluation Results XX
I. Prototype Design (optional, depending on your research type) XX

III
 List of Tables

Table 1. Caption X
Table 2. Caption X
Table 3. Caption X

IV
 List of Figures

Figure 2.1 ESI Triage Algorithm 7

Figure 2.2 Line graph showing prediction ability of the reference model and 9
machine learning models for intensive care use and in-hospital mortality in
the test set
Figure 2.3 Bar chart showing the positive effect of years of experience on E- 11
CTAS usage
13
Figure 2.4 Graph and table showing the operating characteristic curves
15
Figure 2.5 Receiver operating characteristic curve for selected machine
learning models. AUROC: area under the receiver operating characteristic
curve, LR: logistic regression, RF: random forest, XGB: XGBoost
Figure 3.1 Conceptual Framework 17
Figure 3.2 Theoretical Framework 18
21
Figure 4.1 Program Flowchart
22
Figure 4.2 Manual Triage Flowchart

V
VI
 CHAPTER 1

INTRODUCTION

1.1 Background of the Study

Emergency Rooms (ER) have been using a triage system to determine which patients are
to receive urgent treatment or general care services, based on their symptoms. While some
hospitals worldwide still utilize manual registration (otherwise known as written records),
Cicolo and Peres (2019) state that this form of organization is not as accurate or efficient as
electronic documentation when performing the triage. The situation causes overcrowded
waiting rooms - an increasing problem - and results in dissatisfaction for patients (Restrepo-
Zea et al., 2018).

The rising number of patients with dissatisfying experiences and the substandard
satisfaction rate accounted for numerous health care facilities in the Philippines have caused
researchers to wonder what can be changed to help refine the triage process. Anastacia-
Faustino and Ruiz (2018) were determined to alleviate this problem by educating patients and
guardians at the Manila Doctors Hospital of the Emergency Department (ED) processes while
at the triage. Unfortunately, the attempts of improving the system without making an immense
and costly renovation on the part of the hospital management were not successful, as the
additional knowledge of the process did not deter individuals from the evident lack of change
in the system, continuing to cause frustration.

With that being said, the recurrent issue of overcrowded healthcare facilities has caused
numerous researchers to analyze and explore more ways to help improve the system. This
study will add to the current body of knowledge regarding automated triaging systems.

1
1.2 Statement of the Problem

Majority of the hospitals in the Philippines have not adopted an electronic triage system,
leading to the issue of overcapacity in healthcare facilities. The application of
manual/conventional triage systems has been linked with ineffective staff performance,
inaccurate triage decisions, and delayed patient documentation. Decision-making in
emergency rooms can be a matter of life and death, because the wrong order of patient
treatment could easily cost someone a life.

1.3 Research Objectives

This study aims to develop a smart triage system that utilizes Fuzzy Logic to sort patients
based on the Emergency Severity Index (ESI). The following are the specific objectives:

a) To evaluate the accuracy of the algorithm in classifying and queuing patients.

b) Evaluate the overall speed of the system into queuing patients.

c) To develop a graphical user interface that is compatible on a desktop computer.

d) To develop a simple patient database wherein their conditions and basic data are
stored.

1.4 Scope and Limitations

A smart triage system will be developed in order to reduce waiting time on patient sorting
in the Emergency Department. The application will be created solely on the Python
programming language and is therefore only going to be made within its capabilities.

2
 The application will be tested by providing a sample data of patient symptoms and will
be evaluated based on its accuracy in its patient prioritization. However, the application will
not be tested in a hospital setting. Data gathering from actual patients is not included in the
study.

1.4 Significance of the Study

Adapting the system to the management of Philippine hospitals will improve the overall
healthcare system in the ER. This research will help reduce the workload of hospital staff
members by allocating urgent and non-urgent patients in the ER. It will also prevent
overcrowding in the ER by reducing the time that patients have to wait to be treated.

The research will provide more data about the triage systems that are used in the
Philippines. This would bring more knowledge in the academe and help other researchers in
this field of study.

3
 CHAPTER 2

REVIEW OF RELATED LITERATURE

2.1 Introduction

This chapter reviews several research papers done before in regards to emergency rooms,
triage systems, and artificial intelligence.
2.2 Emergency Rooms

Patients with acutely life threatening injuries in the ER need rapid care treatment. The
study made by Frink et al. (2017) presented elements of hospital care for severely injured
patients using publication by Pubmed and German S3 Guidelines. Patients are treated in order
of their prioritization. They are first assessed using Physical Examination and
Ultrasonography, this is according to the Focused Assessment with Sonography in Trauma.
After the assessment, patients are then treated accordingly. The study concluded that severely
injured patients should be treated in a structured fashion such as to the A-B-C-D-E scheme:
securing airway, breathing, and circulation with other whole body examinations.

The emergency room is where patients initially go to, to get their symptoms checked. To
accommodate the patients, emergency room management is imposed. Laue et al. (2019)
provided an extensive overview of the workflow of management, such as priority oriented
treatment algorithms and frequent training of emergency staff for treatments of urgent
patients. The study recognized the frequently used emergency room concept worldwide:
Advanced Trauma Life Support (ATLS) and European Trauma Course (ETC). The ATLS
prioritizes life threatening injuries and considers time as an important factor in this concept.
ETC is training for medical health care professionals to stimulate real life situations in the ER.

One of the biggest problems that all emergency departments have to face is overcrowding.

4
Finding solutions to greatly alleviate this issue can help patients who do require the help they
need to get treatment on time and reduce stress among medical staff members. Bickerdlike et
al. (2015) reports that overcrowding in emergency departments is linked to increased mortality
and diminished quality of care. To minimize this issue, triage scales were found to be very
helpful in identifying least urgent patients and ‘fast tracking’ these patients to their respective
areas of treatment can reduce waiting time in the emergency department. Another possible
way to reduce overcrowding was to involve senior doctors in triaging, however there is not
enough data to recommend transitioning into this form of triaging as there was not enough
substantial data to show the most efficient way this could be implemented.

The emergency department of hospitals in the Philippines is known for having long
waiting times. Jimenez, et al. (2018) aims to determine the factors that lead to the length of
stay in tertiary private hospitals in the Philippines. Results have shown that length of stay is
affected by the patient's age, out of pocket patients, triage levels and time shifts. Older patients
typically tend to have more complications, making treatment and admission longer as
compared to younger patients. Out of pocket patients tend to have difficulty during payment
and funding options, increasing length of stay. Morning and afternoon time shifts were
reported to also have higher length of stay, but more study needs to be done as to why. There
is also reported increased length of stay when there are more urgent and emergent triage cases.

Dealing with overcrowding has become a major issue for most hospitals, and many
researchers have been attempting to find ways to manage it. Anastacia-Faustino and Ruiz
(2018) attempted to alleviate this problem by educating patients and guardians of the processes
in the emergency department. To do this, a group of patients were sorted into two groups, in
which one group was educated on ED processes while the other was not. Both groups would
then be asked to give a patient satisfaction rating. Results show however that there is no
statistical difference between the two groups.

5
 Understanding how the emergency department works is important to better understand
how to mitigate the problems that the ED faces. The research above has shown that an
effective triage system would be beneficial in reducing overcrowding.

2.3 Triage Systems

Different hospitals worldwide have different types of triage system, each with its own set
of advantages and disadvantages. One of the most common triage systems is the sorting of
patients based on a set of colors, with the color green being non-urgent to the red being
extremely urgent. Erenler, et al. (2015) aims to assess the effectiveness of the Green Zone in
emergency departments. The study was done in the Hitit University Çorum Education and
Research Hospital in a span of 3 years. Results show that green zone patients quite frequently
cause overcrowding and a bottle-neck formation at emergency departments, creating
aggression within the patients and the physicians. The lack of medical staff and lack of
measures to prevent non-urgent patients from entering the emergency department is worsening
the overcrowding situation.

One of the most commonly used triage scales is the Emergency Severity Index. Gilboy et
al. (2012) designed an Emergency Severity Index with 5 levels of urgency basing it on four
decision points. Figure 2.1 shows the 4 decision points and the result that each decision may
entail.

6
 Figure 2.1 ESI Triage Algorithm. Reprinted from “Emergency Severity Index (ESI): A Triage Tool for
Emergency Department Care, Version 4,” by Gilboy, N., Tanabe, T., Travers, D., & Rosenau, A. M., 2011,
AHRQ Publication No. 12-0014. Agency for Healthcare Research and Quality.

Another type of triage system is the Manchester Triage System. Its color coding of patients
ranges from blue being non-urgent to red being extremely urgent. Cicolo and Peres (2019)
aims to assess the reliability of the Manchester Triage System in electronic and manual
records. Results have shown that both the usage of electronic and manual recording provide
similar values. However, it has been shown that electronic triage is faster in its triage. This
data shows that an electronic triage system would be more desirable in a hospital granted that

7
the nurses and medical staff know how to use it.

Electronic triage systems often use a form of artificial intelligence in order to sort patients
better. One of the simpler artificial intelligence programs used is fuzzy logic. Dehghani et al.
(2018) designs a triage system based on rule-based reasoning and fuzzy logic based on the
Emergency Severity Index guidelines. Results have shown a 99.44% accuracy rate in terms
of clinical decision support systems. This research will be very helpful in constructing a
similar type of artificial intelligence with fuzzy logic.

Developing emergency department sorting systems that precisely differentiate the

critically sick from stable patients remain hard to accomplish. Machine learning models offer
benefits that can stabilize and improve the predictability of various conditions, yet, no study
has examined the utility of modern machine learning systems. Raita et al. (2019) addresses
the lack of knowledge and developed machine learning models by using available triage data.
The results show that the models showed better performance rather than the more conventional
approach, the Emergency Severity Index (ESI). Implementing modern machine learning
models could help hospitals achieve better clinical care and utilize optimal resources by
enhancing the triage decision making. Figure 2.2 shows that as compared to the reference
model, the machine learning model shows higher net benefit.

8
 Figure 2.2 Line graph showing prediction ability of the reference model and machine learning models for
intensive care use and in-hospital mortality in the test set. Reprinted from “Emergency department triage
prediction of clinical outcomes using machine learning models,” by Raita, Y., Goto, T., Faridi, M. K., Brown,

9
 D. F. M., Camargo, C. A., & Hasegawa, K., 2019, Critical Care, 23(1).

A computerized system of the five level Taiwan triage and acuity scale has been
implemented in Taiwan since 2010. The study conducted by Chang et al. (2016), was to assess
which factors are affecting the critical thinking of the nurses with the newly implemented
computerized system. Content Analysis was used to analyze the interviews and data collected.
Results show the three factors: external environmental, patients’ health status, and nurses’
experience. The study also suggests to ensure the privacy of the patient in triaging, improve
the critical thinking of nurses, and to educate the public about ER visits to improve and make
ER triaging more efficient. Another study conducted by Alumran et al. (2020) analyzed the
Electronic Canadian Triage and the Acuity Scale. Studies resulted in improved accuracy of
the triage systems and the flow of patients in the emergency room. Figure 2.3 shows that
nurses with more years of experience utilize the Electronic Canadian Triage and Acuity Scale
more.

Figure 2.3 Bar chart showing the positive effect of years of experience on E-CTAS usage. Reprinted from

10
“Utilization of an Electronic Triage System by Emergency Department Nurses,” by Alumran, A., Alkhaldi, O.,
Aldroorah, Z., Alsayegh, Z., Alsafwani, F., & Almaghraby, N., 2020, Journal of multidisciplinary healthcare,
13, 339–344.

Based on the information gathered, it has been theorized that the most efficient triage
system would be electronic-based due to its speed and accuracy. Fuzzy logic also seems to be
the best approach for this program. Although there are plenty of types of triage scales that
work similarly, it has been decided that the Emergency Severity Index will be used as the
basis. For the application to work at its best, the nurses would have to have a basic
understanding of how the application works.

2.4 Artificial Intelligence

Numerous factors confirm the survival rate of wounded patients. However, some of the
main factors are the time of sorting and the variety of medical personnel. The ability to
increase the survival rate is limited if the medical staff are only a few. Kim et al. (2018) applied
machine learning algorithms to develop a survival classification system that could
automatically triage patients by using wearable bracelets that do not need to rely on the help
of medical personnel and a classification model that bases its results on patients’ vital signs.
The classification model for survival prediction aims to decrease the time spent on triaging
and increase the survival rate of injured patients in prehospital environments.

Accurately predicting the severity of injured patients is an essential part of the premature
classification of those who are vulnerable. Kang et al. (2020) developed an artificial
intelligence (AI) algorithm to predict the demand for critical care in emergency medical
services (EMS). The algorithm made, used development data collected from the Korean
national emergency department information system during visits in real-time from 151
emergency departments (EDs). Predicted variables for triaging in critical care include: age,
sex, chief complaint, symptom onset to arrival time, trauma, and initial vital signs. The results
showed that the AI algorithm meticulously concluded the need for the critical care of patients

11
 using information during EMS and surpassed the typical triage tools and early warning scores.
Figure 2.4 shows that the electronic triage system outperformed the conventional triage
systems at the area under the receiver operating characteristics.

Figure 2.4 Graph and table showing the operating characteristic curves. Reprinted from “Artificial intelligence
algorithm to predict the need for critical care in prehospital emergency medical services,” by Kang, D.-Y., Cho, K.-

12
J., Kwon, O., Kwon, J., Jeon, K.-H., Park, H., Lee, Y., Park, J., & Oh, B.-H., 2020, Scandinavian Journal of
Trauma, Resuscitation and Emergency Medicine, 28(1).

Recent developments in artificial intelligence (AI) and machine learning algorithms for
numerous applications in emergency medicine have been rapidly expanding. Liu et al. (2018)
classified the study of AI in emergency medicine into three domains, namely AI in predictive
modeling, patient monitoring, and emergency department operations. Some AI application
methods investigated for the research are classification and clustering algorithms, natural
language processing, and text mining. AI will play a valuable role in the emergency
department (ED), in areas such as intelligent monitoring, clinical outcome prediction, and
resource planning. Triage and tracking of patients, as well as results, will be automated, which
will ensure an improved, cost-effective, and efficient emergency medical care.

Some symptoms need to be addressed urgently and Goto et al. (2019) made a specific
technology to help patients with acute coronary syndrome who need urgent revascularization.
An artificial intelligence prediction model that detects data using 12-lead electrocardiogram
for urgent patients who need revascularization in the ER. The model was made to select
patients who need urgent revascularization within 48 hours. to test the effectiveness of the
model, the model was used in the ER of the Keio University Hospital. The artificial
intelligence model was concluded to be effective in producing data and choosing patients who
need the urgent operation.

Artificial intelligence can be helpful in decision making such as doing the triage system.
Farahmand et al. (2017) assessed if using artificial intelligence is accurate when estimating
the emergency severity index version 4 of patients with acute abdominal pain. Patients were
randomly selected for the training. And to compare results of the triage levels, the emergency
medicine physician was used. Results of the assessment show that there is a high level of
accuracy in using artificial intelligence in triaging patients with acute abdominal pain.

Creating an application that utilizes artificial intelligence does not have to be done in such

13
a complex manner, it can be created through the Python programming language. Choi et al.
(2019) developed a machine-learning algorithm based on the Korean Triage and Acuity Scale.
The data used for the research was obtained from the emergency department of a tertiary
hospital and all of the data processing was done with Python. Logistic regression, random
forest, and XGBoost were used to classify data and results have shown that machine learning
models are very effective in ER triaging.

Figure 2.5: Receiver operating characteristic curve for selected machine learning models. AUROC: area
under the receiver operating characteristic curve, LR: logistic regression, RF: random forest, XGB: XGBoost.
“Machine Learning-Based Prediction of Korean Triage and Acuity Scale Level in Emergency Department
Patients,” by Choi, S. W., Ko, T., Hong, K. J., & Kim, K. H., 2019, Healthcare Informatics Research, 25(4),
305.

A simulation model developed by Leiras et al. (2018) helped scale patients with different
symptoms in the hospital laboratory. The model used the Simpy module, a simulation
framework based on the Python programming language. The paper proposed rearranging
patient evaluation into four profiles. The study also wanted to be able to analyze blood

14
sampling and material collecting. Results show that the Simpy model is efficient in helping
the healthcare system. The paper also suggested that only two nurses should perform blood
sampling in the laboratory and that increasing the capacity in the customer register can reduce
patients’ waiting time.

The Python programming language is beneficial in hospitals, especially during the

COVID-19 pandemic. Presented in a paper done by McCombs and Kadelka (2020), the model
explored the effects of policy choices in three domains: social distancing, hospital triaging,
and testing in SARS-CoV-2 and COVID-19. The study aims to categorize the population into
two: high-risk and low-risk, and modeled distinctions in social interactions due to context, risk
level, infection status, and testing status. The model, which was implemented and tested on
Python 3.7, used several significant characteristics of the disease to understand the outbreak
better. The results support current policies to contain the breakout, suggest possible
refinements to public health policy and education, and provide a viable explanation as to why
some countries are more successful in suppressing the disease.

Artificial Intelligence has proved to be a very efficient substitute for manual triaging.
Numerous results have shown that accuracy, increased survival rate, improved emergency
medical care, and effective triaging are the most common outcomes in this kind of research.
Based on the data collected, artificial intelligence could be the most beneficial and impressive
replacement for medical personnel and manual triage assessment.

15
 CHAPTER 3

THEORETICAL AND CONCEPTUAL FRAMEWORK

3.1 Introduction

This chapter provides the conceptual framework of the application. Theoretical

considerations are also discussed in this chapter.
3.2 Conceptual Framework

Figure 3.1. Conceptual Framework

The basis of the triaging system application in sorting its patients is from the Emergency

16
 Severity Index (ESI). The source as to which the triage information was obtained is from the
book: Emergency Severity Index (ESI): A Triage Tool for Emergency Department Care,
Version 4. (Gilboy et al., 2012).
3.3 Theoretical Considerations

Figure 3.2. Theoretical Framework

Although the triage scale used in this triage system is the Emergency Severity Index, other
triage scales with proven reliability can also be taken into consideration. The Manchester
triage system and Five-level Taiwan triage and acuity scale have all shown good results from
previous studies.

17
 CHAPTER 4

METHODOLOGY

4.1 Materials

As stated on the website page of GCF Global, a computer is an electronic device that
manipulates information or data. It can store, retrieve, and process data. These devices come
in a variety of types. For this paper, the application has been made applicable on desktop
computers as they are more common and convenient inside hospitals. However, during the
duration of this study, laptop computers were also utilized in making and testing the
application.

4.2 Procedure
1. Gather data for training and testing of the algorithm.

Practice cases and standards of ESI triage index levels will be gathered by the
researchers to use in training the algorithm and database of the application. Practice cases,
programs, and official standards of the ESI will be collected from an ESI handbook vers.
4 by Gilboy et al. (2012), and practice cases from the Emergency Nurses Association. The
handbook of the ESI will be used as the foundation of the decision-making process of the
application. Practice cases will act as a patient’s data to test the application’s decision
making. All data will be imputed in the system, to test and train the ability of the
application as a whole. These will make sure that the application is accurate in decision
making and processing the patient’s symptoms and needs.

2. Create the Software GUI.

18
 A theme or aesthetic will be chosen for the GUI. The agreed design will be created
using python toolkit, Tkinter. Structures and widgets provided by the toolkit will be added
to the application.

3. Implement a database program integrated into the GUI.

A simple database will be created and easily integrated into Python. Practice
patient’s data and personal information will be inputted here. The practice cases will test
how the application will react with stored data. The database will be combined with the
GUI to implement a style in the whole application.

4. Implement the fuzzy logic decision algorithm of the ESI of the Patient.

Fuzzy logic decision algorithms will be implemented using Scikit-Fuzzy Toolkit.

This concept will be responsible for the decision-making process of the application. ESI
will be implemented in this part of the application due to it being the basis of the decision
process.

5. Test the accuracy of the algorithm

Accuracy of the algorithm will be examined by calculating the accuracy rate. To

compute for the accuracy, the error rate and accuracy rate should be first evaluated. The
accuracy rate of the application should be ideally above 80%.

6. Test the speed of manual and electronic triage systems.

The time spent by the test subjects in using the manual triage system and the
electronic triage system will be compared and evaluated. The algorithm flowchart is seen

19
in Figure 4.1. and while the manual triage flowchart is seen in Figure 4.2.

Figure 4.1. Program Flowchart

20
 Figure 4.2. Manual Triage Flowchart

4.3 Research Design

1. Data gathering for the AI (training data)
a. Emergency Severity Index
i. Factors indicated on the index

The guidelines used for this project are based on Gilboy et al.
(2012) version of the Emergency Severity Index (ESI). The factors that
decide a patient’s severity level include:

21
 - The need for life-saving interventions
- Patient rating of pain/distress from 1-10
- Resources needed
- Vital signs: Body temperature, pulse rate, blood pressure and
respiration rate
ii. ESI definition

The Emergency Severity Index (ESI) is a five-level triage scale

used to rapidly assess patients’ urgency based on their conditions. This
scale is one of the different types of triaging systems used worldwide. The
ESI does not only evaluate the patient's level of urgency but it also presents
the resources that are needed for the patient's care. All patients are
evaluated according to their symptoms and criteria that will display their
level of urgency.

iii. Classification

The ESI version 4 triage system evaluates the patient using a five-
level scale. Level 1 patients, usually have an unstable condition, should be
immediately evaluated and treated. In terms of resources, due to them
being a high-risk patient they would need multiple resources depending on
their affliction. Level 2 patients do not belong in level 1 but should also be
urgently treated, should be sent to an acute treatment area for them to be
monitored. Level 3 patients, usually confused with their mental status,
should be sent to a minor trauma area. Level 3 patients are predicted to
only need 3 resources to treat their ailment. The lower levels, levels 4 and
5 are patients with acute or minor trauma. They are predicted to only use

22
 less than two resources to treat them.

b. Data Preparation

The database creates a compilation of practice patient’s personal

information including name, date of birth. address, contact details, gender,
pre-existing medical conditions, allergies, physician, and dates of
discharge. Training data will be supplemented before the test set will be
used to evaluate the triage system.

2. Algorithm and Application

As compared to Boolean Logic, the truth values in Fuzzy Logic may range
from integers 0 to 1 as there is no completely right or wrong answer. The
algorithms that would be used are from SciKit-Fuzzy toolkit. An example of the
antecedents includes:

● pain level
○ Universe (ie, crisp value range): Rate the pain/distress level from
1-10.
○ Fuzzy set (ie, fuzzy value range): low, medium, high

An example of the consequents includes:

● level
○ Universe (ie, crisp value range): Rate the urgency of the patient
from 1-5.
○ Fuzzy set (ie, fuzzy value range): 1, 2, 3, 4, 5

23
 An example of the rules includes:

● IF the pain level was high, THEN the level will be 4.

a. Look into how to create GUI in Python

Graphical User Interface (GUI) is a system, it uses objects for users to

interact with the information. According to Banerjee I. (2017), The object focuses
on designing the look of the application that would be made. To create GUI in
Python Tkinter, a toolkit in python, can be used as an easy way to create GUI. As
specified by Coders Legacy (2020), a Tkinter window should be outlined by its
frame to add structure, and then Tkinter Widgets would be added. Basic Tkinter
Widgets include Buttons, Listbox, Label, and Entry. Tkinter GUI Widgets include
Check Button, Radio Button, and Scrollbar. Tkinter Menu Widgets include Menu,
Combobox, and Menu button. Advanced Tkinter Widgets includes Canvas, Top
level, and File Dialogs. And Tkinter Scale Widgets Includes a Scale.

b. Database in python

The triage system will keep a patient database for medical health personnel
use. Text files and excel files will be used to store patient information. An example
of how the database would look is seen in table 4.1.

Table 4.1. Patient Database

c. Target results and what the results imply (i.e.: 50% accurate, what does it mean)
● Accuracy of the Algorithm

24
 ○ The triage data will be used to evaluate the accuracy and error rate
of the triage system. The accuracy and the error rate will be
determined by the following equations:

𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦
𝑇𝑟𝑢𝑒 𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑅𝑎𝑡𝑒 + 𝑇𝑟𝑢𝑒 𝑁𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑅𝑎𝑡𝑒
= 𝑥 100
𝐴𝑙𝑙 𝑆𝑎𝑚𝑝𝑙𝑒𝑠

𝐸𝑟𝑟𝑜𝑟 𝑟𝑎𝑡𝑒
𝐹𝑎𝑙𝑠𝑒 𝑁𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑅𝑎𝑡𝑒 + 𝑇𝑟𝑢𝑒 𝑁𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑅𝑎𝑡𝑒
= 𝑥 100
𝐴𝑙𝑙 𝑆𝑎𝑚𝑝𝑙𝑒𝑠

● Speed of Using the Triage vs Manual

○ The speed between the analysis speed of the algorithm and the
manual triage system will be evaluated by determining whether
there is a statistically significant difference between the two.
● Fuzzification and Defuzzification of Data
○ Fuzzification is the transformation of crisp value inputs into fuzzy
input sets. Then, by applying the rules, the output can obtain the
fuzzy output set. Defuzzification then changes this into crisp value
output.
4.4 Data Analysis Strategy

To determine if there is a significant difference between the analysis speed of the algorithm
and manual triage systems, there will be a given set amount of 10 patients (controlled
variable). One of the students will try to do the manual paper-based test while another would
perform the simulation. The final time to finish the 10 patients will be gathered and recorded.
It will be repeated into 10 trials. Z-Test statistical test will be used if the changes between the

25
time of the automated and manual are statistically significant. The hypothesis is as follows:

H10: There is a statistically significant difference in the rate of algorithm and manual triage
system.

H1a: There is no difference in the rate of algorithm and manual triage system.

To determine the accuracy of the algorithm and manual triage systems, the
correctness/accuracy of the results and the number of patients completed will be evaluated
based on a given set amount of time. The results would be the ratio of the total number of
patents correct vs wrong on each of the methods. The resulting ratios will then undergo Z-test
treatment to check if the difference is statistically significant. The hypothesis is as follows:

H10: There is a statistically significant difference in the accuracy and completed patients
of algorithm and manual triage system.

H1a: There is no difference in the accuracy and completed patients of algorithm and
manual triage system.

The Z-test that will be used for both data is a two-tailed test of two proportions.

26
 REFERENCES

Alumran, A., Alkhaldi, O., Aldroorah, Z., Alsayegh, Z., Alsafwani, F., & Almaghraby,
N. (2020). Utilization of an Electronic Triage System by Emergency Department
Nurses. Journal of multidisciplinary healthcare, 13, 339–344.
https://doi.org/10.2147/JMDH.S250962

Anastacia-Faustino, J. A. & Ruiz, R. L. (2018). Effect of patient education in the

triage in improving patient satisfaction in the Emergency Department of
Manila Doctors Hospital. The Filipino Family Physician, 56(2), 52-57.
http://www.herdin.ph/index.php?view=research&cid=69128#apa

Banerjee I. (2017, January 19). Based Testing of GUI-Based Software. Chapter 2,

Advances in model. Doi:10.1016/bs.adcom.2016.12.003

Bickerdike, L., Wilson, P., & Booth, A. (2015). Triage and minimising crowding
in emergency departments. University of York.
https://www.york.ac.uk/crd/publications/effectiveness-matters/triage-
emergency-crowding/

Chang, W., Liu, H. E., Goopy, S., Chen, L. C., Chen, H. J., & Han, C. Y. (2017).
Using the Five-Level Taiwan Triage and Acuity Scale Computerized
System: Factors in Decision Making by Emergency Department Triage
Nurses. Clinical Nursing Research, 26(5), 651–666.
https://doi.org/10.1177/1054773816636360

Choi, S. W., Ko, T., Hong, K. J., & Kim, K. H. (2019). Machine Learning-Based
Prediction of Korean Triage and Acuity Scale Level in Emergency
Department Patients. Healthcare Informatics Research, 25(4), 305.
doi:10.4258/hir.2019.25.4.305

Cicolo, E. A., & Peres, H. H. (2019). Electronic and manual registration of

Manchester System: Reliability, accuracy, and time evaluation. Revista
Latino-Americana De Enfermagem, 27. doi:10.1590/1518-

27
 8345.3170.3241

Coders Legacy. (2020). Python GUI. Python. Retrieved from

https://coderslegacy.com/python/python-gui/

Dehghani Soufi, M., Samad-Soltani, T., Shams Vahdati, S., & Rezaei-Hachesu, P.
(2018). Decision support system for triage management: A hybrid
approach using rule-based reasoning and fuzzy logic. International
journal of medical informatics, 114, 35–44.
https://doi.org/10.1016/j.ijmedinf.2018.03.008

Emergency Nurses Association. 2020. Emergency Severity (ESI) Resources and

FAQs. https://www.ena.org/education/esi

Erenler, A. K., Özel, I., Ece, Y., Karabulut, M., Oruçoğlu, A., &amp; Çiftçi, E.
(2015). Analysis of Triage Application in Emergency Department. Open
Journal of Emergency Medicine, 03(03), 13-17.
doi:10.4236/ojem.2015.33003

Farahmand, S., Shabestari, O., Pakrah, M., Hossein-Nejad, H., Arbab, M., &
Bagheri-Hariri, S. (2017). Artificial Intelligence-Based Triage for Patients
with Acute Abdominal Pain in Emergency Department; a Diagnostic
Accuracy Study. Advanced Journal of Emergency Medicine, 1(1), e5.
doi:10.22114/AJEM.v1i1.11

Frink, M., Lechler, P., Debus, F., & Ruchholtz, S. (2017, July 24). Multiple
trauma and emergency room management. Deutsches Arzteblatt
International. Deutscher Arzte-Verlag GmbH.
doi:10.3238/arztebl.2017.0497

GCF Global. Computer Basics: What is a Computer?

https://edu.gcfglobal.org/en/computerbasics/what-is-a-computer/1/.

Gilboy, N., Tanabe, T., Travers, D., & Rosenau, A. M. (2011) Emergency

28
 Severity Index (ESI): A Triage Tool for Emergency Department Care,
Version 4. Implementation Handbook 2012 Edition. AHRQ Publication
No. 12-0014. Agency for Healthcare Research and Quality.

Goto, S., Kimura, M., Katsumata, Y., Goto, S., Kamatani, T., Ichihara, G., …
Sano, M. (2019). Artificial intelligence to predict needs for urgent
revascularization from 12-leads electrocardiography in emergency
patients. PLoS ONE, 14(1). doi:10.1371/journal.pone.0210103

Grinberg, M. (2018). Flask web development: developing web applications with

python. " O'Reilly Media, Inc.".

Jimenez, D., Manzanera, R., Mira, J., Navarro, I., & Wong, J. (2018). A
Descriptive Study on the Factors Affecting the Length of Stay in the
Emergency Department of a Tertiary Private Hospital in the Philippines.
Acta medica Philippina. 52. 521-526.

Kang, D.-Y., Cho, K.-J., Kwon, O., Kwon, J., Jeon, K.-H., Park, H., Lee, Y.,
Park, J., & Oh, B.-H. (2020). Artificial intelligence algorithm to predict
the need for critical care in prehospital emergency medical services.
Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine,
28(1). https://doi.org/10.1186/s13049-020-0713-4

Kim, D., You, S., So, S., Lee, J., Yook, S., Jang, D. P., Kim, I. Y., Park, E., Cho,
K., Cha, W. C., Shin, D. W., Cho, B. H., & Park, H.-K. (2018). A data-
driven artificial intelligence model for remote triage in the prehospital
environment. PLOS ONE, 13(10).
https://doi.org/10.1371/journal.pone.0206006

Lansdowne, K., Scully, C. G., Galeotti, L., Schwartz, S., Marcozzi, D., & Strauss,
D. G. (2015). Recent Advances in Medical Device Triage Technologies
for Chemical, Biological, Radiological, and Nuclear Events. Prehospital
and Disaster Medicine, 30(3), 320–323.
doi:10.1017/S1049023X15004641

29
Laue, F., Ramadanov, N., & Matthes, G. (2019). Emergency room management
for severely injured patients. Notfall Und Rettungsmedizin, 22(1), 63–78.
doi:10.1007/s10049-018-0552-2

Leiras, A., González-Calderón, C. A., de Brito Junior, I., Villa, S., & Yoshizaki,
H. T. Y. (Eds.). (2019). Operations management for social good: 2018
poms international conference in Rio. Springer Nature.

Liu, N., Zhang, Z., Wah Ho, A. F., & Ong, M. E. H. (2018). Artificial intelligence
in emergency medicine. Journal of Emergency and Critical Care
Medicine, 2, 82. https://doi.org/10.21037/jeccm.2018.10.08

McCombs, A., & Kadelka, C. (2020). A model-based evaluation of the efficacy of

COVID-19 social distancing, testing and hospital triage policies.
doi:10.1101/2020.04.20.20073213

Oltrogge, J., Brandt, B., & Mayer-Runge, U. (2019). Scherer MUniversity

hospital emergency department: Nurse versus primary care physician
triage. Zeitschrift fur Allgemeinmedizin,(6) 252-258.
doi:10.3238/zfa.2019.0252–0258

Raita, Y., Goto, T., Faridi, M. K., Brown, D. F. M., Camargo, C. A., &
Hasegawa, K. (2019). Emergency department triage prediction of clinical
outcomes using machine learning models. Critical Care, 23(1).
doi:10.1186/s13054-019-2351-7

Restrepo-Zea, J. H., Jaén-Posada, J. S., Piedrahita, J. J. E., & Flórez, P. A. Z.

(2018). Emergency department overcrowding: A four-hospital analysis in
Medellín and a strategy simulation. Revista Gerencia y Politicas de Salud,
17(34). https://doi.org/10.11144/Javeriana.rgps17-34.ssua

Soufi, M. D., Samad-Soltani, T., Vahdati, S. S., &amp; Rezaei-Hachesu, P.

(2018). Decision support system for triage management: A hybrid

30
approach using rule-based reasoning and fuzzy logic. International
Journal of Medical Informatics, 114, 35-44. doi:10.1016/j.ijmedinf.2018

31
32