A Web Mobile Decision Support System To Improve Medical Diagnosis Using A Combination Of K-Mean And Fuzzy Logic

TELKOMNIKA, Vol.17, No.6, December 2019, pp.3145~3154
ISSN: 1693-6930, accredited First Grade by Kemenristekdikti, Decree No: 21/E/KPT/2018
DOI: 10.12928/TELKOMNIKA.v17i6.12715 ◼ 3145
Received March 21, 2019; Revised July 2, 2019; Accepted July 18, 2019
A web/mobile decision support system to improve
medical diagnosis using a combination of
K-Mean and fuzzy logic
Zainab T. Al-Ars*1
, Abbas Al-Bakry2
1
University of Information Technology and Communication,
Department of Remote Sensing and GIS, College of science, University of Baghdad, Iraq
2
University of Information Technology and Communications Baghdad, Iraq
*Corresponding author, e-mail: zainabdrweesh@scbaghdad.edu.iq
Abstract
This research provides a system that integrates the work of data mining and expert
system for different tasks in the process of medical diagnosis, and provides detailed steps to
the process of reaching a diagnosis based on the described symptoms and mapping them with
existing diagnosis available on the web or on a cloud of medical knowledge based, aggregate
these data in a fuzzy manner and produce a satisfactory diagnosis of the persisting problem.
The mobile phone interface would make the system user-friendly and provides mobility and
accessibility to the user, while posting updates and reading in details the steps that led to
the decision or diagnosis that is reached by the K-mean and the fuzzy logic inference engine.
The achieved results indicate a promising diagnosis performance of the system as it achieved
90% accuracy and 92.9% F-Score.
Keywords: AI, communication, expert system, fuzzy logic, K-Mean clustering
Copyright © 2019 Universitas Ahmad Dahlan. All rights reserved.
1. Introduction
Health care fields have found their way among those fields to make use of computer
sciences. The intersection between computer sciences and health care has been formulated
into a new filed of studies called “medical informatics”, where medical data are collected, stored,
processed, analyzed, retrieved, and used in various medical related operations. Different
disciplines have emerged to cover the wide range of specialties required by the medical
informatics field. These include [1]:
- Bioinformatics: workers in this field are concerned with collecting, storing, retrieving and
analyzing medical data either for research purposes or to provide better patients’ care.
- Public health informatics: workers of this area define the way public can make use of medical
records and what records are available to researchers and medical practitioners.
- Electronic Health Records: these records are exchanged between different healthcare
providers to provide better care for patients, and workers of this field are responsible for
securing this exchange, and allowing only authorized personnel have access.
- Health Data Analyst: this type of analysts uses medical data to define trends and relations
between different health records to come up with predictions and recommendations to some
medical situations.
The medical domain is one of the most important and critical domains that has direct
interaction with the humans’ wellbeing. Some wrongly diagnosed illness or badly made
decisions can have a serious effect on someone’s health or life. Errors in diagnosis can be
related to different reasons, like the lack of experience in the healthcare provider, inaccurate
description of the symptoms, or the lack of information available for the medical staff. Wrong
decisions could originate from lack of cooperation between different healthcare providing
department or hesitance in making the decision. Medical institutes are in constant need for
computerized help to provide better diagnosis to certain illnesses, whether for collecting large
amounts of data, analyzing complex input, organizing and classifying data, finding relations

◼ ISSN: 1693-6930
TELKOMNIKA Vol. 17, No. 6, December 2019: 3145-3154
3146
between these data and many other operations that when computerized would provide great
help in the diagnosis process.
One of the areas where medical informatics show its great influence on health care is
the help in providing diagnosis of diseases according to some rules that are put to work on huge
amounts of medical records of patients and medical sciences. Sometimes, these diagnoses can
help in accelerating providing required medical assistant faster, which would help improve
the quality of life for human beings. Medical decision making process involves different actions
to be taken before reaching a satisfactory decision that can help in improving patients’ care
outcomes, like diagnosis, prognosis, treatment, and therapeutic monitoring [2].
Fuzzy logic is an approach to computing and problem solving that provides “degrees of
true” not the common binary true or false solutions. Decision making systems that are based on
fuzzy logic resemble the way a human being makes decisions, by having levels of truthiness,
according to [3]. Medical diagnosis system has a series of fuzzy data as an input to
the diagnosis process, where a fuzzy logic based decision making system need to be adopted
to deal with these “Shades” of correct and wrong decisions [4]. Classifying these data in fuzzy
sets is necessary in the process of diagnosis to deal with different levels of uncertainty in
the production of the final diagnosis (decision).
The widespread of mobile and wearable devices provides an opportunity and also
concern in the process of collecting, analyzing, and transmission of medical data. Opportunities
shown in different aspects, most important ones show in the time efficiency of data collection,
the flexibility of patient’s movement and personal time management, in addition to being easily
adaptable to existing mobile application installed on smart devices (like smart mobile phones).
2. Related Work
Diagnosis of a disease or a medical care is looked at as a decision making process that
is based on the collected data and the relations that could be conducted from the integration of
these data. Since these data are mostly in natural language and can be explained with some
degree of uncertainty, which affects the classification on which the rules used to make a
decision depends, most researchers in Decision Support Systems (DSS) generally prefer
the use of fuzzy logic techniques.
Fuzzy logic classification defines sets of data based on a threshold of provided values.
If the developed defines this threshold ahead of running the fuzzy logic mechanism, then
the sets are considered to be of type-1 fuzzy sets, where the intervention of the developer is
needed. While in other systems the threshold is adapted from the fuzzificaiton process and can
be dynamically changes throughout the course of the fuzzification process, so the resulting sets
are considered to be of type-2 fuzzy sets [3]. Due to the huge amounts of data and the wide
variations of their values and sources, recent researchers focus on adopting type-2 sets in
the fuzzy logic process, as in the work in [5], where type-2 fuzzy sets were used in an
automated decision making system for home health care of diabetes management of home
treated patients.
Another research adopted fuzzy decision making technique in the selection of
vaccination application of a healthcare system [6]. Choosing the suitable vaccine to admit to
patients is challenging and the decision depends on linguistic variables provided by physicians
to rate the alternatives. The inference engine was provided with Fuzzy distance measure for all
alternatives of the training sets (varied between people, spatial and temporal data) along with
crisp values of fuzzy weights. Testing the system showed that People and temporal data
provided to the DSS were the most suitable vaccination method for protecting people from
H1N1 influenza epidemic (the test case).
In home treated patients, the decision making efficiency and speed could be the divider
between good life and serious medical issues. The authors of [7] adopted a fuzzy logic DSS in
Personal Home Healthcare System for Cardiac Patients. Data are collected from the patients
through sensors (could be fuzzy or crisp data), where are used as input to the fuzzifier and
whereby the membership functions defined these data are applied to their actual values
(membership function) to help set a degree of correctness in the rules base. With the wide
spread of fuzzy data, other types of fuzzy sets were developed like intuitionistic fuzzy set, type-n
fuzzy set, fuzzy multisets, and hesitant fuzzy set [4]. Researchers had attempts towards
benefiting the integration between these types of sets (in addition to the mostly used

TELKOMNIKA ISSN: 1693-6930 ◼
A web/mobile decision support system to improve medical diagnosis using… (Zainab T. Al-Ars)
3147
type-2 sets) as in the work in [4] where hesitant decisions were supported through
the integration of type-2 fuzzy set, hesitant sets and intuitionistic fuzzy when having doubts
between different values to be classified.
Samuel et al introduced a Web-Based Decision Support Sys-tem (WBDSS) coupled
with fuzzy logic (FL) for typhoid fe-ver (TF) diagnosis. The KB of the system contained a fuzzy
inference system (FIS). The system was developed to aid in the provision of accurate and timely
TF diagnosis. Studies on the proposed system were performed based on the medical records of
TF patients. The efficiency of the proposed system was based on the standard statistical
metrics, while the achieved results showed a 94% efficiency of the system in providing an
accurate diagnosis. The authors suggested the integration of ANN into the FL-based medical
diagnosis sys-tems for better performances [8].
Fatumo et al designed a diagnostic ES called XpertMalTyph for the diagnosis of
different types of typhoid and malaria complications. The ES simulates the skills of the medical
expert in disease diagnosis using computers. Hence, the ES can provide similar services in
the absence of an expert, making it possible to treat patients even from their homes. They
suggest the implementation of the ES with artificial neural networks (ANN). The XpertMalTyph
was executed in Java Expert System Shell [9].
3. System Architecture and Work Flow
The system allows the user to register information and the system will diagnose
the patient’s condition and find out biological problem and its cause depending on a database
that is stored online. System makes judges depending on AI program built on fuzzy logic.
The database information is captured depending on data acquisition unit which gets new
information only from users who input the diagnostics to database after visiting physician.
The novelty in this system will be in providing user with full description on how did it
reach to the result, like writing down in English what steps it took to find out what's wrong. This
is the major data acquisition unit job which includes filtering information and causes and
providing description for each. Such an intelligent system and with enough database will be able
to reply with meaningful description with humans normally via text.
System will include an android application that will be able to listen to humans
describing problems, convert speech to text then skimming text for useful information that will be
used to build diagnostics, then convert the reply from text to speech and say it to patient. Such
an interactive system will be an amazing result of applying AI in useful biomedical apps.
The hybrid system is implemented over 2 phases: phase 1 involved clustering of data collected
through data acquisition unit, and phase 2 is the expert system that holds the AI module
(fuzzy logic).
3.1. Data Clustering
An intelligent system requires huge amounts of data to build its knowledge and learn or
make decisions according to these data and the learnt patterns or relationships between them.
An efficient clustering mechanism is required to make sure that these huge amounts of data are
usable and would enhance the learning and decision-making processes. These “big data”,
especially in the healthcare industry, are changing the way patients and doctors handle care.
The bigger data involved, the more efficient healthcare services are, yet the harder it is to
manage these data.
A cluster holds a collection of data items that are aggregated together based on some
similarities between them. K-means clustering is one of the simplest and most popular learning
algorithms, due to its simplicity and low computational costs. Clustering of data using k-means
starts with a first group of randomly selected centroids, which are used as the starting points for
every cluster, and then performs iterative calculations to optimize the positions of the centroids,
and the data items that are similar to that centroid [10]. Clusters creation stops when:
− The centroids have stabilized i.e. (There is no change in their values because the clustering
has been successful).
− The defined number of iterations has been achieved. This number is defined by
the programmer or set based on certain threshold of the number of clusters.
The Κ-means clustering algorithm uses iterative refinement to produce a final result.
The algorithm inputs are the number of clusters Κ and the data set. The data set is a collection

◼ ISSN: 1693-6930
3148
of features for each data point. The algorithm starts with initial estimates for the Κ centroids,
which can either be randomly generated or randomly selected from the data set. The algorithm
then iterates between two steps [11]:
1. Data assignment step:
Each centroid defines one of the clusters. In this step, each data point is assigned to its
nearest centroid, based on the squared Euclidean distance. More formally, if ci is the collection
of centroids in set C, then each data point x is assigned to a cluster based on
𝑆𝑖 = 𝑐𝑖 ∈ 𝐶
⏞
𝑎𝑔𝑟 𝑚𝑖𝑛
𝑑𝑖𝑠𝑡(𝑐𝑖. 𝑥)2
(1)
where dist (·) is the standard (L2) Euclidean distance. And for each ith cluster centroid the set of
data point assignments is Si.
2. Centroid update step:
In this step, the centroids are recomputed. This is done by taking the mean of all data
points assigned to that centroid's cluster.
𝑐𝑖 =
1
|𝑆𝑖|
∑ ∈ 𝑆𝑖
𝑥𝑖
𝑥𝑖
(2)
the algorithm iterates between steps one and two until a stopping criterion is met (i.e., no data
points change clusters, the sum of the distances is minimized, or some maximum number of
iterations is reached).
This algorithm is guaranteed to converge to a result. The result may be a local optimum
(i.e. not necessarily the best possible outcome), meaning that assessing more than one run of
the algorithm with randomized starting centroids may give a better outcome. Choosing K
(denoting the number of clusters) is considered to be the backbone of the algorithm to run as
desired. To find the number of clusters in the data, the user needs to run the K-means clustering
algorithm for a range of K values and compare the results. In general, there is no method for
determining exact value of K, an estimation is performed using some tested technique. One of
the metrics that is commonly used to compare results across different values of K is the mean
distance between data points and their cluster centroid. Since increasing the number of clusters
will always reduce the distance to data points, increasing K will always decrease this
metric [12-15]. Mean distance to the centroid as a function of K is plotted and the "elbow point"
where the rate of decrease sharply shifts, can be used to roughly determine K [16].
A number of other techniques exist for validating K, including cross-validation,
information criteria, the information theoretic jump method, the silhouette method, and
the G-means algorithm. In addition, monitoring the distribution of data points across groups
provides insight into how the algorithm is splitting the data for each K. The proposed hybrid
system is dedicated to use the k-means algorithm in clustering of collected data that are used to
get the proper medical diagnosis, based on symptoms submitted by the user. The clustering
model building process is shown in Figure 1.
Figure 1. Phase 1 of the system’s building (Build a clustering Model)

3149
3.2. The Expert System
An Expert System is one of the most common applications of artificial intelligence. It is a
computer program that simulates the decision and actions of a person or an association that
has specialist facts and experience in a particular field [17]. Normally, such a system contains a
knowledge base containing accumulated experience and a set of rules for applying
the knowledge base to each particular situation. The major features of expert system are user
interface, data representation, inference, explanations etc. Advantages of expert system are
increased reliability, reduced errors, reduced cost, multiple expertise, intelligent database,
reduced danger etc. Disadvantages of expert system are absence of common sense and no
change with changing environment [18, 19].
A Fuzzy Expert System, that is a decision-making system based on Fuzzy Logic
inference, is a group of membership functions and rules. These functions and rules are used to
reason about data. Fuzzy expert systems are oriented toward numerical processing. It takes
numbers as input, and then translates the input numbers into linguistic terms like Small, Medium
and large. Then the task of Rules is to map the input linguistic terms onto similar linguistic terms
describing the output. Finally, the translation of output linguistic terms into an output number is
done [20]. These rules are built in an if-then manner, and are evaluated in parallel in
the inference engine, which indicates that the orders of these rules are not important. The terms
used to explain the rules are defined ahead of initializing the inference engine. These terms are
nouns and adjectives that are used to describe the input data (like high, normal, big, low,
etc.) [21, 22].
Fuzzy logic relies on having learning rules, according to which an inference engine
makes decisions regarding the problem at hand. The input to these fuzzy sets is in natural
language (mostly) and the output can either be crisp or in natural language [3]. A fuzzy classifier
is a procedure of labeling sets in the decision making or machine learning algorithm that
embeds uncertainty measures, aka; fuzzy logic, within its workflow. A fuzzy classifier uses a rule
base that can be expressed as a fuzzy knowledge base to Convert crisp input into a linguistic
variable in a process named fuzzififcation, then an inference engine makes fuzzy decisions
based on pre-set rules, and the fuzzy output is defuzzified by converting it back into crisp output
using membership functions analogous to the ones used in the fuzzification phase [22].
The analysis of fuzzy input in order to produce a fuzzy decision is based on three
main operations:
− Receive the fuzzy input: that could be from a single source, multiple heterogonous sources
or even distributed data sources.
− Processing of these fuzzy inputs in a “fuzzification” technique that relies on a set of rules that
are set according to human thinking “if-then” procedures in simple natural language, in
addition to traditional processing methods.
− Reaching weighted results after passing the fuzzy rules and assembling them into a single
decision related to the problem that guides other parts of the system or the human user what
to do after de-fuzzification of the results to make them understandable.
The expert system that is built using a fuzzy logic inference module uses the clustered
data produced by the data collection unit using the k-means algorithm to build the fuzzy
knowledge base. Once the user enters his/her data then the symptoms he/she suffers from,
the system matches these symptoms to a suitable cluster. This cluster is determined by a fuzzy
logic system. Another fuzzy logic module uses patient’s attributes (age, environment, gender)
along with the cluster’s attributes and features extracted from the symptoms, to make a decision
on which diagnosis is the most relevant. Figure 2 shows the main structure of the expert
system’s functionality.
The fuzzy logic module produces a ranked list of matching diagnosis. Each diagnosis is
given a confidence level in percentage. Based on this level, the system chooses the one with
the highest level and sends it to the user’s device. If this confidence level is low (below 60%)
then the system prompts the user either to enter more symptoms or to suggest adding
the diagnosis (assuming that a physician is using the system or the patient has consulted a
doctor to make a judgment). The user sends this suggested diagnosis back to the expert
system, which uses this diagnosis to alter the rules, or add a new rule to cover this diagnosis
and learn from it to produce a more reliable diagnosis for similar cases in the future.

◼ ISSN: 1693-6930
3150
Figure 2. Expert System's main structure
4. System Evaluation
The system was tested by a group of experts in the field of medical diseases diagnosis
and found to give a high performance and accurate results together with the ease of use.
Central Pediatric Teaching Hospital, World Health Organization (Factsheet), WebMD,
Mayoclinic, healthline, and other certificated website were used for data collection. The medical
dataset of 350 records (diseases) and 300 features (symptoms, disease description and
the possible advice and treatment according to the disease degree where collected, analyzed,
and preprocessed to the required format [23].
4.1.The Fuzzy Expert System
The fuzzy logic toolbox is one of most powerful tools that MATLAB provides, to help its
users build, analyze, design and simulate a fuzzy logic system, through a set of applications that
encapsulate functions easily used through its user-friendly interface and code integration
module. The functions fuzzy logic toolbox provides cover the basic methods used by fuzzy logic,
like fuzzy clustering, neuro-fuzzy learning, inference engine and rule base building. The rules
are implemented using simple logic rules and integrated within the fuzzy inference system,
which can be later used to simulate the fuzzy system as a while, or simulate it within Simulink
(a part of Matlalb’s environment) to simulate the fuzzy system within a comprehensive model of
the entire dynamic system. The data acquisition unit collected and clustered big data regarding
common diseases and related symptoms. These data are used to build rules to be integrated
into the rule base of the fuzzy logic to decide on what is the most relevant disease the submitted
symptoms show. An example of such a rule is:
− Disease (Patient, tuberculosis),
− Symptom (Patient, persistent_cough),
− Symptom (Patient, constant_fatigue),
− Symptom (Patient, weight_loss),
− Symptom (Patient, loss_of_appetite),
− Symptom (Patient, fever),
− Symptom (Patient, coughing_up_blood),
− Symptom (Patient, night_sweats).
where Tuberculosis is a lung disease whose symptoms are persistent cough, constant fatigue,
weight loss, loss of appetite, fever, coughing up blood, night sweats.
The expert system takes into consideration other factors when building a diagnosis. As
it is known to almost every healthcare practitioner; factors like age, environment, and previous
health issues do affect the decision made about diagnosis. The fuzzy logic based expert
systems, uses as input, in addition to the submitted symptoms, user’s age, medication history,
Body Mass Indicator, gender. These attributes are submitted by the user through the diagnosis
unit, and used by the diagnosis unit to reach a decision. The fuzzy logic engine accepts
non-crisp input of normalized data as a “Fuzzy” input for the inference engine to make a “fuzzy”
decision. In other words; the fuzzy logic evaluation system takes input values between 0-1 by

3151
mapping each of the “fuzzy set” members through a membership function so that all values are
between 0 and 1. The performance of our proposed system was evaluated using a confusion
matrix which contains the information on the actual case of the patient diagnosed by the expert
and the diagnosis predicted by our hybrid system. A two-class classifier confusion matrix is
shown in Table 1 and the result is presented in Table 2.
Table 1. Confusion Matrix
Confusion Matrix
Predicted
Negative Positive
Actual
Negative
TN: is the number of diagnosed cases
(No, they don't have a disease)
predicted correctly but they don't have
the disease
FP: is the number of diagnosed cases
(Yes, they do have a disease) predicted
incorrectly but they do not have the
disease.
Positive
FN: is the number of diagnosed cases
(No, they don't have a disease)
predicted incorrectly and they do have
the disease.
TP: is the number of diagnosed cases
(Yes, they do have a disease) predicted
correctly and they do have the disease.
Table 2. The Proposed System Confusion Matrix
Proposed System Confusion Matrix
Predicted
Total
Incorrect by the FRS Correct by the FRS
Actual
Incorrectly diagnosis cases 6 1 7
Correctly diagnosis cases 2 21 23
The performance of our proposed system was generally rated using the data in
the matrix. Some metrics, including accuracy, precision, sensitivity (recall), F-measure
(F1 score), and specificity were applied as the criteria to implement this evaluation. In (3) to (7)
show the formulas for these metrics [24]:
𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 =
𝑇𝑁 + 𝑇𝑃
𝑇𝑁+𝐹𝑃+𝐹𝑁+𝑇𝑃
× 100% (3)
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 =
𝑇𝑃
𝑇𝑃+𝐹𝑃
× 100% (4)
𝑅𝑒𝑐𝑎𝑙𝑙 =
𝑇𝑃
𝑇𝑃+𝐹𝑁
× 100% (5)
𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐𝑖𝑡𝑦 =
𝑇𝑁
𝑇𝑁+𝐹𝑃
× 100% (6)
𝐹1 − 𝑆𝑐𝑜𝑟𝑒 = 2 ×
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 × 𝑅𝑒𝑐𝑎𝑙𝑙
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛+ 𝑅𝑒𝑐𝑎𝑙𝑙
(7)
The F1-Score is the harmonic average of the precision and recall. The best value
realized is one and the worst is zero [25]. As shown in Table 3, the F1 score of our system was
0.929 which indicates a very good performance.
Table 3. The Accuracy of the Proposed System
Results
Accuracy 90%
Precision 95%
Sensitivity (Recall) 91%
Specificity 85%
F1-Score 0.929
4.2.User Interface
The system was implemented to be compatible with a website browser where the user
can input the symptoms in the front end as a textual expression; these expressions are matched
against the closest symptom collected by the data collection unit. The user can also choose
the symptom from a list of pre-defined symptoms. The screenshot in Figure 3 shows
the system’s start up page, where the user enters personal attributes to be used as input
(in addition to the symptoms) to the fuzzy logic expert system.

◼ ISSN: 1693-6930
3152
The user enters his/her information and moves to the next page to start adding
the sympotoms he/she has to help the system provide proper diagnosis. Figure 4 shows
the symptoms entry web page. The symptoms entered by the user can be given a level of
effect. How annoying or uncomfortable these symptoms are making the user. These levels are:
Low (somewhat annoying), Moderate (annoying and causes pain sometimes), and High (this
symptom is obstructing the user from performing daily chores). When the user finishes adding
the symptoms and clicks the “finish” button, the system produces the diagnosis that produced
the highest “confidence level” in the fuzzy logic expert system. This diagnosis is displayed with
how sever it is (low, moderate, or high) along with proposed procedure to follow or a treatment.
Figure 5 shows the diagnosis web page.
The user is able to enter symptoms in a similar manner to that through a web browser
adding symptoms is easy through the drop-down list or by simply filling in the text box. Choosing
the severity of the symptoms is done through radio buttons for easy and clear input. Once
the symptoms are entered, the diagnosis window is displayed.This information is sent to
the base station to the main diagnosis unit to add to the confidence level produced by the expert
system to make future diagnosis even more accurate, and provide future users with a more
reliable diagnosis system, and automatically modify the rules in the knowledge base of
the system, and modify the learning and inference rules accordingly too.
Figure 3. Start up web page for the proposed medical diagnosis system
Figure 4. Entering symptoms

3153
Figure 5. Diagnosis based on submitted symptoms
Figure 6. Part of add diagnosis web page
Figure 7. Android emulator operating the diagnosis system
5. Conclusion
An expert system that is built with a Fuzzy Logic decision making procedure is a big
step towards having and adaptive and constantly developing system that builds its own
knowledge base and provide users with an adequate diagnosis procedure that supports levels
of severity of detected diseases, based on several information, among which are the symptoms

◼ ISSN: 1693-6930
3154
provided by the system’s users themselves. The use of clustering technique increases
the system performance by groubing the most similer diseases which allows the fuzzy logic to
look up for the right diagnosis in the nearest cluster according to Euclidean distance.
The achieved results indicate a promising diagnosis performance of the system as it
achieved90% accuracy and 92.9% F-Score.
References
[1] Parker KR, Srinivasan SS, Houghton RF, Kordzadeh N, Bozan K, Ottaway T, Davey B.
Health informatics program design and outcomes: Learning from an early offering at a mid-level
university. Education and Information Technologies. 2017; 22(4): 1497-1513.
[2] Shen Y, Colloc J, Jacquet-Andrieu A, Lei K. Emerging medical informatics with case-based reasoning
for aiding clinical decision in multi-agent system. Journal of Biomedical Informatics. 2015; 56:
307-317.
[3] Bellman RE, Zadeh LA. Decision-making in a fuzzy environment. Management Science.1970; 17(4):
B-141.
[4] Xia M, Xu Z. Hesitant fuzzy information aggregation in decision making. International Journal of
Approximate Reasoning. 2011; 52(3): 395-407.
[5] Gatton TM, Lee M. Fuzzy logic decision making for an intelligent home healthcare system. In 2010 5th
International Conference on Future Information Technology.2010.
[6] Lopez D, Gunasekaran M. Assessment of vaccination strategies using fuzzy multi-criteria decision
making. In Proceedings of the Fifth International Conference on Fuzzy and Neuro Computing
(FANCCO-2015). Springer, Cham. 2015; 195-208.
[7] Hussain A, Wenbi R, Xiaosong Z, Hongyang W, da Silva AL. Personal home healthcare system for
the cardiac patient of smart city using fuzzy logic. Journal of Advances in Information
Technology.2016; 7(1): 58-64.
[8] Samuel O, Omisore M, Ojokoh B. A Web Based Decision Support System driven by Fuzzy Logic for
the diagnosis of typhoid fever. Expert Systems with Applications. 2015; 40:4164-417.
[9] Fatumo S, Adetiba E, Onolapo J. Implementation of XpertMalTyph: An Expert System for Medical
Diagnosis of the Complications of Malaria and Typhoid. IOSR Journal of Computer Engineering
(IOSR-JCE). 2013; 8(5): 34–40.
[10] Page JT, Liechty ZS, Huynh MD, Udall JA. BamBam: genome sequence analysis tools for biologists.
BMC Research Notes.2014; 7(1): 829.
[11] Hartigan JA, MA Wong. A K-means clustering algorithm. Applied Statistics. 1979; 28:100–108.
[12] F Yuan Z, H Meng, HX Zhang, CR Dong. A New Algorithm to Get the Initial Centroids. Proc. of the
3rd International Conference on Machine Learning and Cybernetics, 2004; 26-29.
[13] Z Huang. Extensions to the k-means algorithm for clustering large data sets with categorical values.
Data Mining and Knowledge Discovery. 1998; 2: 283-304.
[14] D Birant A. Kut. ST-DBSCAN: An algorithm for clustering spatial-temporal data. Data & Knowledge
Engineering. 2007; 60(1): 208-221.
[15] ALN Fred, JMN Leitão. Partitional vs hierarchical clustering using a minimum grammar complexity
approach. Proc. of the SSPR & SPR 2000. LNCS 1876. 2000; 193-202.
[16] K-means algorithm in Python tutorial, available through:
https://pythonprogramminglanguage.com/kmeans-elbow-method/ accessed, Nov. 2018.
[17] Bassem S. Medical Expert Systems Survey. International Journal of Engineering and Information
Systems (IJEAIS). 2017; 1(7): 218-224.
[18] De Kock E. Decentralising the Codification of Rules in A Decision Support Expert Knowledge Base,
M.Sc. Thesis. Faculty of Engineering. Built Environment and Information Technology, University of
Pretoria; 2003.
[19] Jimmy S, Dinesh G, Abhinav B. Medical Expert Systems for Diagnosis of Various Diseases.
International Journal of Computer Applications. 2014; 93(7): 36-43.
[20] Imène B, Noria T. A multi-agent framework for a web-based decision support system applied to
manufacturing system. CIIA. 2009; 9.
[21] William S, James J. Fuzzy Expert Systems and Fuzzy Reasoning, John Wiley & Sons, Inc., 2005.
[22] Rimpy N. Medical Expert System-A Comprehensive Review. International Journal of Computer
Applications. 2015; 130(7): 44-50.
[23] Centers for Medicare & Medicaid Services. ICD-10-CM Official Guidelines for Coding and Reporting.
2012 ICD-10-CM and GEMs. Retrieved May 17, 2012.
[24] Zainab T. Al-Ars and Abbass Al-Bakry. Iraq's Major Infectious Disease Diagnosis Using A Fuzzy
Rule-Based System. International Journal of Engineering & Technology .2018; 7(4): 4943-4948.
[25] Chala D, Million M, Debela T. Developing a Knowledge-Based System for Diagnosis and Treatment
of Malaria. Journal of Information & Knowledge Management. 2016; 15(4): 108-112.

A Web Mobile Decision Support System To Improve Medical Diagnosis Using A Combination Of K-Mean And Fuzzy Logic

More Related Content

Similar to A Web Mobile Decision Support System To Improve Medical Diagnosis Using A Combination Of K-Mean And Fuzzy Logic (20)

More from Wendy Berg (20)

Recently uploaded (20)

A Web Mobile Decision Support System To Improve Medical Diagnosis Using A Combination Of K-Mean And Fuzzy Logic