Ijp volume 4 issue 3_pages 1465-1473

Int J Pediatr, Vol.4, N.3, Serial No.27, Mar 2016 1465
Original Article (Pages: 1465-1473)
http:// ijp.mums.ac.ir
Evaluation of Suspected Pediatric Appendicitis with Alvarado
Method Using a Computerized Intelligent Model
*Amir Jamshidnezhad11
, Ahmad Azizi 1
, Saeed Shirali 2
, Sara Rekabeslamizadeh1
, Maryam
Haddadzadeh 3
, Yalda Sabaghan 4
1
Department of Health Information Technology, Faculty of Allied Health Sciences, Ahvaz Jundishapur
University of Medical Sciences, Ahvaz, Iran.
2
Department of Laboratory Sciences, Faculty of Allied Health Sciences, Ahvaz Jundishapur University of
Medical Sciences, Ahvaz, Iran.
3
Faculty of Medicine, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran.
4
Ahvaz Imam Educational Hospital, Ahvaz, Iran.
Abstract
Background: Acute appendicitis is one of the common and urgent illnesses among children. Children
usually are unable to help the physicians completely due to weakness in describing the medical
history. Moreover, acute appendicitis overlaps with conditions of other diseases in terms of Symptoms
and signs in the first hours of presentation. These conditions lead to unwanted biases as well as errors
for diagnosis of acute appendicitis and increase the medical costs for hospitals and patients. The
purpose of this study is to develop a computer based model in the diagnosis of Acute appendicitis
specially at pediatrics field.
Materials and Methods: Fuzzy-rule based systems are popular methods widely used in the Clinical
Decision Support Systems (CDSSs). In this article, a hybrid Fuzzy rule based system as a CDSS was
compared with the Alvarado method for diagnosis of appendicitis in children. In this system an
evolutionary algorithm was also developed to create and optimize the Fuzzy rules for diagnosis of
pediatrics. To find the performance of the proposed model, a dataset was created from children with
abdominal pain who referred to the teaching general hospitals in Ahvaz, Iran in 2013 to 2014 years. In
this process, the results achieved from the developed model were compared with the Alvarado scoring
system used for children with abdominal pain in the previous studies.
Results: The experimental results showed that the developed model has a proper performance to
detect the patients with acute appendicitis from others techniques and models such as the Alvarado
scoring system.
Conclusion: The developed model can be used as an assistant for physicians in the diagnosis of acute
appendicitis, especially in the field of pediatrics.
Key Words: Appendicitis, Fuzzy logic, Meta-heuristic algorithms, Pediatric.
*Please cite this article as: Jamshidnezhad A, Azizi A, Shirali S, Rekabeslamizadeh S, Haddadzadeh M,
Sabaghan Y. Evaluation of Suspected Pediatric Appendicitis with Alvarado Method Using a Computerized
Intelligent Model. . Int J Pediatr 2016; 4(3): 1465-73.
*Corresponding Author:
Amir Jamshidnezhad, Faculty of Allied Health Sciences, Ahvaz Jundishapur University of Medical Sciences,
Ahvaz, Iran.
Email: a.jamshidnejad@yahoo.com
Received date Jan 10, 2016 ; Accepted date: Feb 22, 2016

Alvarado Method in Pediatric Appendicitis
1. INTRODUCTION
Clinical diagnosis is affected with
various amounts of clinical symptoms,
biological data and available clinical tests
(1). The diagnostic challenge of some
diseases is appeared when they have
similar conditions in terms of symptoms,
risk factors and experimental data (2).
Acute appendicitis as one of the high risk
acute abdomen presents the similar
symptoms of other disorders (3). The risk
is dramatically increased for children
under six years old. Clinical decision to
diagnosis of appendicitis is often made
based on physicians’ intuition and
experience. Its overlapped symptoms lead
to un-wanted biases as well as errors and
medical costs for hospitals and patients.
Overall, diagnostic accuracy of acute
appendicitis ranges from 58% to 92%
which shows the higher risk in female
patients (4). Therefore, the clinical
decision support systems could decrease
these diagnostic errors as well as unwanted
practice variation and improves the patient
treatments in terms of times and costs (5).
Alvarado score test as a clinical scoring
system is a diagnostic tool for analyzing
the signs and symptoms of acute
appendicitis (16). However, inappropriate
of sensitivity was demonstrated for
diagnosis of appendicitis in the case of
pediatrics. This method has developed in
the further studies to improve the accuracy
rate for detection of acute appendicitis (7-
9). In recent years, Fuzzy logic has been
used to model the problems with
uncertainty domains for the purpose of
making more accurate and real results.
Fuzzy logic with using fuzzy sets defines
mathematically intermediate values
between crisp values such as black and
white to apply human thinking way in the
computer systems. Therefore, fuzzy logic
can be used to formulate the imprecise
problems such as decision making,
classification, pattern recognition, control,
and etc. Fuzzy Rule Based Systems
(FRBS) are based on the IF-Then rules
which contain linguistic variables to use
the human knowledge in the intelligent
systems. Nowadays, fuzzy logic as well as
other artificial intelligence techniques, are
used extensively in the Clinical decision
support systems (CDSSs). Neural networks
are the popular methods used in the
clinical diagnostic systems as the learning
techniques. Recently, neural networks
were used in the fuzzy detection systems
to assist the physicians to increase the
accuracy of acute appendicitis diagnosis
(1). The outcomes showed that the hybrid
Fuzzy-neural networks systems can be
used as an effective solution for prediction
of acute appendicitis (1, 10, 11). However,
current studies have still restrictions to
model accurately a diagnostic system
while limited clinical items and laboratory
measurements are available. Moreover,
many computational or optimization
problems are not solved simply with using
classical mathematical techniques as they
need to handle a lot of complex solutions.
Therefore, sometimes the results will be
local optima rather than global optima
(12). On the other hand, the traditional
methods can be used for just a specific
problem and are not enough intelligent to
use in every conditions as a whole
solution.
Genetic Algorithms (GAs) are a popular
computational technique that follows the
natural process of genetics to find the
global optima solution among the huge
number of solutions (13). Holland is well
known to introduce and develop genetic
algorithms to solve the complex problems
based on the evolutionary computation
(14). The idea of genetic algorithms has
been captured from the Charles Darwin’s
theory of natural selection that creatures
follow for survivals. Genetic algorithms
as a popular method of evolutionary
computation are used to solve the
optimization problems based on the
evolutionary process. Therefore, genetic

Jamshidnezhad et al.
algorithms with the search in the solutions
space find the best or approximately
results in a process of several generations
(15). Nowadays, numerous applications for
genetic algorithms are known in the
different areas such as engineering,
management, medicine, humanity sciences
and computer science (16).
Recently conducted research has however
simulated the honey bee offspring
generation process to enhance the GAs
performance for optimizing the detection
problems (17). The simulated honey bee
algorithm called Bees Royalty Offspring
Algorithm (BROA) improves
considerably, the accuracy of the classical
GAs in the complicated fields (18). BROA
formulates the biological process of honey
bee offspring generation which leads with
the Queen bee in the colony. BROA
customized to use for the diagnostic
system developed in this article. BROA as
a meta-heuristic algorithm have the
optimizing role to tune the knowledge base
of the proposed fuzzy system. The meta-
heuristic/fuzzy system is thus proposed to
classify the patients with Right Iliac Fossa
(RIF) pain into appendicitis and non-
appendicitis groups. In this model, BROA
was used to create the rules in order to
optimize the fuzzy knowledge base. Then,
the results of the diagnostic model were
compared with ultrasound imaging
findings to show the performance of the
developed system. In the next sections the
proposed hybrid model as well as its parts
and performance for the purpose of acute
appendicitis diagnosis are described in
details.
1.1 Clinical aspects of Acute
Appendicitis
Appendicitis is the most common
abdominal emergency. The life-time risk
of developing appendicitis is around 7%
from 10 to 30 year-old patients which
usually requires surgery (19). Generally,
the diagnostic error in the appendectomy
patients is approximately between 8% -
45% varies based on the gender (20, 21).
The most important method to confirm a
diagnosis of the acute appendicitis is the
laboratory tests while the diagnostic
sonography procedures can improve the
accuracy of appendicitis prediction.
Various pediatrics studies showed the
sensitivity and specificity of 85-97 and 47-
96 respectively, for diagnosis of acute
appendicitis. There are several clinical and
biochemical factors considered with the
physicians to decide about patients who
are suffering with the abdominal pain.
Table.1 shows the most important factors
usually are monitored by the doctors for
diagnostic and treatment of acute
appendicitis. However, high risk of
infection and life lost in delayed diagnosis,
variety of signs and overlapped symptoms
with other diseases as well as the time
consuming, high costs and variable
availability of imaging procedures are the
main reasons to show that computer based
models may be a valuable diagnostic aid
when appendicitis is suspected to be the
underlying cause of an acute abdomen,
particularly in low-resource countries,
where imaging is not available option (2).
2. MATERIALS AND METHODS
In this research, a Hybrid Fuzzy
Genetic Algorithm based model was the
proposed as a differential diagnosis
technique. The model was optimized while
the knowledge base is created and tuned
with the proposed algorithm (17). The
input parameters for the proposed meta-
heuristic-fuzzy classifier system are age,
first abdominal pain time, initial pain site,
Right Lower Quadrant (RLQ) abdomen
shift, white blood cell (WBC) count and
neutrophil count. Clearly, the selected
features were a few of signs and symptoms
showed in the (Table.1). Therefore, the
proposed model was aimed to diagnose the
appendicitis with a few features to
decrease the limitation of children
diagnosis. Overall, for the purpose of

increasing the accuracy of the differential
diagnosis in the appendicitis pediatric, a
computer based model was developed in
which a few parameters were used as the
input factors. The process of classification
was performed using a modified genetic
algorithm simulated from honey bees
mating process called Bee Royalty
Offspring Algorithm (BROA), as an
optimization method to fit the fuzzy
knowledge with the problem conditions.
Figure.1, shows the architecture of the
proposed diagnostic classification system.
Table 1: Anthropometry, physical
examinations and laboratory factors for
diagnosing of acute appendicitis
Factors Factors
Age Previous surgery
Gender Pain shift
Pain site Foetor
Pain nature Change of micturition
Pain time RIF tenderness
Nausea or vomiting Rebound tenderness
Rigidity Guarding
Temperature Temperature
White blood cell count
(WBC)
Neutrophil count
Figure 1: Architecture of the proposed differentiate diagnosis system
2-1. Abdominal data-set
To evaluate the accuracy of the proposed
model for diagnosing of appendicitis, a
collection of data was gathered from
abdominal patients with the age under 15
years who referred to the Educational
hospitals of Ahvaz, a metropolis in South
West of Iran, in 2013 to 2014 years. This
dataset included some features such as
patient’s signs and symptoms, biochemical
tests as well as physicians’ first diagnosis
and final diagnosis about the acute
appendicitis. Only 6 features were used to
evaluate the reliability and validity of the
model in the experimental process.
Therefore, two training and testing set
were randomly selected from the dataset.
To show the performance of the developed
model, the obtained results were compared
with the outcomes achieved from the
Alvarado scoring system as a popular
scoring system to diagnosis of acute
appendicitis.
2-2. Fuzzy Knowledge base
The computational procedure of fuzzy
inference system based on the knowledge
base to predict the new conclusion is
implemented in the inference engine.
Therefore, knowledge of fuzzy diagnostic
model was determined and optimized in

the knowledge base component by the
BROA to find the best rules set for
diagnosis of appendicitis. In this model the
rules were trained to classify the input data
into three classes which showed the
Presence, Absence and High risk
diagnostic conditions. Figure 2, shows the
general frame-work of fuzzy knowledge
base including rules, features (fi) and the
acceptable values. The matrix as the input
vector was optimized by the proposed
meta-heuristic algorithm for diagnosis of
acute-appendicitis.
Figure 2: The overall structure of Fuzzy knowledge base as the input vector of the meta-heuristic
algorithm
2-3. Defuzzification
The process of converting fuzzy output
values to a crisp value was determined
according to the mean value of maximum
(MOM) defuzzification strategy.
2-4. Meta-heuristic optimizing
algorithm
For the purpose of improving the
performance of the fuzzy diagnostic
system, a genetic-based algorithm called
BROA as a Meta-heuristic technique was
used. Genetic algorithms are the popular
methods to solve the optimizing problems.
In the current study, BROA was proposed
to create and optimize the fuzzy
knowledge base. Therefore, although the
fuzzy rules are normally set by the experts,
due to diversity of features as well as their
effects on the diagnosis of appendicitis, the
rules were created in a training process.
Moreover, the achieved results were also
compared with the Alvarado findings
while the data of patients with acute
abdomen suspicious of acute appendicitis
from hospitals in Iran were used.
2-5. Ethics
The ethics approval to conduct this
research was granted by the Research
Ethics and Review Committee of the
Ahvaz Jundishapur University of Medical
Sciences, Iran.
3. RESULTS
Table 2, shows the overall outcomes
from the proposed system on the testing
set. The achieved results are also
compared with Alvarado test as well as
other techniques in (Table3).
According to Table 2, the highest
predication rate was related to the Reject
status. Therefore, the model has the best
performance to diagnose of the patients
suffering with other diseases. However,
the accuracy rate for the abdominal
patients who were confirmed finally to get
the appendicitis was 64%. Similarly, only
50% of the high risk patients were
predicted accurately in the proper class.
Overall the accuracy rate to classify the
abdominal patients into three categories
was around 71.4%.

Table 2: Diagnosis rate for acute appendicitis in children
Diagnostic status Overall Accuracy rate
Presence 64%
High Risk 50%
Reject 100%
Average 71.4%
Table 3: Comparison of the diagnostic models for acute appendicitis prediction
Diagnostic model Age category Number of used factors Accuracy rate
Alvarado Clinical Scoring System
(ACSS) (26)
All ages 9 66.4%
Modified Alvarado System (30) Children Not mentioned 56%
Alvarado System (24) Children 8 66.9%
Modified Alvarado System (25) Children 7 64%
Multi Layer Neural Networks (27). All ages 8 92.89%
SVM (28) All ages 10 99%
Neural Networks (29) All ages 10 90%
Neural Networks (26) All ages 9 74%
Neural networks(11) All ages 11 97%
Proposed model Children 6 71.4%
4. DISCUSSION
According to experimental results, the
average accuracy of the proposed
differential diagnosis model for acute
appendicitis in population was around 72
percent. The result was achieved with 6
input factors including: age range, first
abdominal pain time, initial pain site,
Right Lower Quadrant (RLQ) abdomen
shift, white blood cell (WBC) count and
neutrophil counts. To show the
performance of the model, existing
methods such as Alvarado scoring system
were compared and considered as follows:
4-1. Alvarado Scoring System
Alvarado scoring system is widely used as
a decision support system for diagnosis of
acute appendicitis in recent years.
Influence of Alvarado on clinical decision
making in acute appendicitis was
determined in several studies (22-26, 30).
In this study, the achieved results from
Alvarado scoring system on suspected
appendicitis in Iran was compared with the
results from the proposed computer based
model. Findings showed that the Modified
Alvarado method has the sensitivity,
specificity and overall accuracy of 53.9, 74
and 56 percent, respectively for children in
Iran. The results illustrated that the
Alvarado for diagnosis of pediatric
appendicitis is not as reliable as diagnosis
in adults (25).
Moreover, in the diagnostic process of
Alvarado method more features were used
as input data in comparison with the
hybrid model developed in this study.
According to (Table.3) the best Alvarado
performance showed less than 68% of
accuracy rate for diagnosis of appendicitis.
Generally, the diagnosis of acute
appendicitis in children population showed
less accuracy rate in comparison with other
age categories. Therefore, the Alvarado

method showed higher error of diagnosis
while needs to some input data which are
difficult to determined according to
children medical history. On the other
hand, the proposed model in this study
illustrated better performance in
comparison with the current method.
4-2. Other Computer- based diagnostic
techniques
According to Table.3, neural network is a
popular technique to classify the presence
of acute appendicitis from its absence for
the data of the abdominal patients.
However, neural networks needed to use
large size of features including clinical and
nonclinical examination factors as the
input variables to show the accurate
performance. Neural networks are not
robust when the input factors are less than
10 features. Moreover, the Alvarado
scoring system as the most cited model is
not as valid as other techniques for acute
appendicitis prediction (26).
On the other hand, although the Support
Vector machines (SVM) showed a good
performance in detection of acute
appendicitis, it needs to at least 10 factors
to classify the patients with acute
appendicitis from non-acute appendicitis
(27). The proposed model including meta-
heuristic algorithm in this study resulted a
performance with 90% accuracy. Although
the achieved result in this study was lower
than the accuracy of some algorithms used
in the existing research, it used only 6
input factors for prediction. Moreover, the
model analyzed the data from abdominal
patients into three classes including
presence and absence of acute appendicitis
or high risk for acute appendicitis.
Moreover, with regarding to the fact that
determining larger number of the input
factors needs to more laboratory tests or
nonclinical examination, time costing is
one of the weak points of the existing
research in comparison with the proposed
model. The model also showed a valuable
performance in comparison with
physician’s primary diagnosis for acute
appendicitis. According to the reported
diagnosis for acute appendicitis in the
dataset, the presence of acute appendicitis
for 23% of patients with acute appendicitis
was rejected in the primary diagnosis
while the acute appendicitis was found in
the final diagnosis. Therefore the primary
diagnosis by physicians showed around
77% accuracy for accepting the acute
appendicitis.
5. CONCLUSION
Medical diagnosis in terms of time and
accuracy has the vital role to patients’ life.
In this study a hybrid decision support
system was used to diagnosis of acute
appendicitis in children. The proposed
optimizing algorithm increased the
accuracy of prediction with improving the
knowledge base in the fuzzy rule based
system. The results showed a proper
performance especially in reject status of
the acute appendicitis. The diagnostic
acceptable performance was achieved
while a limited range of factors as the
input parameters were used in the hybrid
model.
To compare the developed model
performance, the diagnosis results from
Alvarado scoring system used for
suspected appendicitis in children were
considered. The comparison results
showed the better performance of
developed hybrid model in comparison
with Alvarado scoring system. Therefore,
the model improves the processing time as
well as the treatment costs to diagnosis of
acute appendicitis in children. As a result,
the computer- based model can be used as
an assistant for physicians to making
decision in the urgent condition for
differentiate diagnosis of appendicitis in
children especially conditions in which
access to costly equipment such as CT
scans and sonography tools are limited.

6- CONFLICT OF INTEREST: None.
7-ACKNOWLEDGMENT
This research was supported by Ahvaz
Jundishapur university of Medical
sciences. Authors thank our colleagues
from Ahvaz Imam Educational hospital
that greatly assisted the research for
preparing the data set.
8-REFERENCES
1. Sivasankar E, Rajesh RS,
Venkateshwaran S.R. Appendicitis diagnosis
system using fuzzy logic- and neural network-
based classifier. Int J of Med Eng and Info
2011; 3(4): 337-48.
2. Ohle R, O'Reilly F, O'Brien K, Fahey
T, Dimitrov BD. The Alvarado scores for
predicting acute appendicitis: a systematic
review. BMC Medicine 2011; 9:139.
3. Bundy DG, Byerley JS, Liles EA,
Perrin EM, Katznelson J, Rice HE. Does this
child have appendicitis. JAMA 2007; 298(4):
438-51.
4. Petroianu A. Diagnosis of acute
appendicitis. Int J of Surgery 2012; 10(3):115–
19.
5. Wu R, Peters W, Morgan MW. The
Next Generation Clinical Decision Support:
Linking Evidence to Best Practice. Journal of
Health care Information Management 2002;
16(4): 50-5.
6. Alvarado A. A practical score for the
early diagnosis of acute appendicitis. Ann
Emerg Med 1986; 15: 557-64.
7. Gwynn LK. The diagnosis of acute
appendicitis: clinical assessment versus
computed tomography evaluation. J Emerg
Med 2001; 21: 119-23.
8. Shepherd J. The use of the clinical
scoring system by Alvarado in the decision to
perform computed tomography for acute
appendicitis in the ED. Am J Emerg Med
2007; 25: 489-93.
9. Chan MY, Teo BS, Ng BL. The
Alvarado score and acute appendicitis. Ann
Acad Med Singapore 2001; 30: 510-12.
10. Tez M, Tez S, Go¨cmen. E.
Neurofuzzy is Useful Aid in Diagnosing Acute
Appendicitis. World J of Surg 2008; 32: 2126.
11. Prabhudesai S, Gould S, Rekhraj
S,Tekkis P, Glazer G, Ziprin P. Artificial
neural networks: useful aid in diagnosing acute
appendicitis. World J. Surg 2008; 32: 305-9.
12. Bonissone P, Khedkar P, Chen YT.
Genetic algorithms for automated tuning of
fuzzy controllers: a transportation application,
Proc. of the IEEE Conference on Fuzzy
Systems (FUZZ-IEEE ‘96), New Orleans,
Louisiana. p. 674–80.
13. Mitchell M. An Introduction to
Genetic Algorithms Complex Adaptive
Systems. Bradford Books. MIT Press; 1996.
14. Holland JH. Adaptation in Natural and
Artificial Systems: An Introductory Analysis
with Applications to Biology, Control, and
Artificial Intelligence. University of Michigan
Press; 1975.
15. Sivanandam SN, Deepa SN.
Introduction to Genetic Algorithms. Springer;
2007. ISBN: 354073189X, 9783540731894.
16. Coley DA. An Introduction to Genetic
Algorithms for Scientists and Engineers.
Wspc; Har/Dskt editionWorld: Scientific;
1999. ISBN-10: 9810236026.
17. Jamshidnezhad A, Nordin JMD. A
heuristic model for optimizing fuzzy
knowledge base in a pattern recognition
system. J of Sci & Indust Research 2012; 71:
341-47.
18. Jamshidnezhad A, Nordin JMD. Bee
royalty offspring algorithm for improvement
of facial expressions classification model. Int J
of Bio Inspired Computation 2013; 5(4):175-
91.
19. Rosengren D, Brown AF, Chu K.
Radiological imaging to improve the
emergency department diagnosis of acute
appendicitis. Emerg Med Australas 2004; 16:
pp. 410-16.
20. Raman SS, Lu DS, Kadell BM,
Vodopich DJ, Sayre J, Cryer H. Accuracy of
nonfocused helical CT for the diagnosis of
acute appendicitis: a 5-year review. AJR Am J
Roentgenol 2002; 178: 1319-25.
21. Wong SK, Chan LP, Yeo A. Helical
CT imaging of clinically suspected
appendicitis: correlation of CT and histological
findings. Clin Radiol 2002; 57: 741-45.
22. Saidi RF, Ghasemi M. Role of
Alvarado Score in diagnosis and treatment of
suspected acute appendicitis. Am J Emerg
Med 2000; 18(2): 230-31.

23. Macklin CP, Radcliffe GS, Merei JM,
Stringer MD. A prospective evaluation of the
modified Alvarado score for acute appendicitis
in children. Ann R Coll Surg Engl 1997;
79(3): 203-5.
24. Al Hashemy AM, Seleem MI.
Appraisal of the modified Alvarado Score for
acute appendicitis in adults. Saudi Med J 2004;
25: 1229-31.
25. Yeganeh R, Peivandi H, HajiNasrolahi
E, Salehi N, Ahmadi M, Bidardel F. A
prospective evaluation of the modified
Alvarado score for acute appendicitis in
children. J of Med Council of Iran 2008; 26(4):
515-21.
26. Sakai S, Kobayashi K, Nakamura J,
Toyabe S, Akazawa K. Accuracy in the
27. diagnostic prediction of acute
appendicitis based on the Bayesian network
model. Methods Inf Med 2007; 46: 723-726.
28. Park SY, Kim SM. The Optimization
Variables of Input Data of Artificial Neural
Networks for Diagnosing Acute Appendicitis.
Apld Math & Info Sci 2014; 8(1): 339-43.
29. Park SY, Seo JS, Lee SC, Kim SM.
Application of an Artificial Intelligence
Method for Diagnosing Acute Appendicitis:
The Support Vector Machine. Future Info
Tech. Lecture Notes in Electrical Engineering
2014; 276: 85-92.
30. Hsieh CH, Lu RH, Lee NH, Chiu WT,
Hsu MH, Li YC. Novel solutions for an old
disease: diagnosis of acute appendicitis with
random forest, support vector machines and
artificial neural networks. Surgery 2011;149:
87-93.
31. Duglass S. Diagnosis of acute
Appendicitis in children using a clinical
Practical practice Guidline. J of pediatric surg
2004; 39(3): 458-63.

Ijp volume 4 issue 3_pages 1465-1473

More Related Content

What's hot (18)

Similar to Ijp volume 4 issue 3_pages 1465-1473 (17)

Recently uploaded (20)

Ijp volume 4 issue 3_pages 1465-1473