Automated diagnosis of hepatitis b using multilayer mamdani

Research Article
Automated Diagnosis of Hepatitis B Using Multilayer Mamdani
Fuzzy Inference System
Gulzar Ahmad ,1
Muhammad Adnan Khan ,1
Sagheer Abbas ,1
Atifa Athar,2
Bilal Shoaib Khan,3
and Muhammad Shoukat Aslam1
1
Department of Computer Science, National College of Business Administration and Economics, Lahore, Pakistan
2
Department of Computer Science, CUI, Lahore, Pakistan
3
Department of Computer Science, Minhaj University, Lahore, Pakistan
Correspondence should be addressed to Gulzar Ahmad; gulzar.phd@ncbae.edu.pk
Received 25 June 2018; Accepted 3 December 2018; Published 5 February 2019
Academic Editor: Emanuele Rizzuto
Copyright © 2019 Gulzar Ahmad et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In this research, a new multilayered mamdani fuzzy inference system (Ml-MFIS) is proposed to diagnose hepatitis B. The
proposed automated diagnosis of hepatitis B using multilayer mamdani fuzzy inference system (ADHB-ML-MFIS) expert system
can classify the different stages of hepatitis B such as no hepatitis, acute HBV, or chronic HBV. The expert system has two input
variables at layer I and seven input variables at layer II. At layer I, input variables are ALTand ASTthat detect the output condition
of the liver to be normal or to have hepatitis or infection and/or other problems. The further input variables at layer II are HBsAg,
anti-HBsAg, anti-HBcAg, anti-HBcAg-IgM, HBeAg, anti-HBeAg, and HBV-DNA that determine the output condition of
hepatitis such as no hepatitis, acute hepatitis, or chronic hepatitis and other reasons that arise due to enzyme vaccination or due to
previous hepatitis infection. This paper presents an analysis of the results accurately using the proposed ADHB-ML-MFIS expert
system to model the complex hepatitis B processes with the medical expert opinion that is collected from the Pathology De-
partment of Shalamar Hospital, Lahore, Pakistan. The overall accuracy of the proposed ADHB-ML-MFIS expert system is 92.2%.
1. Introduction
Disease analysis is a crucial element in the field of medicine
and healthcare. An inappropriate analysis of a disease often
results in improper treatment that leads to complications of
the ailment and eventually to death [1]. What are the major
signs and symptoms of the disease and its extent or degree
of symptoms on the organs? When this is resolved, suitable
treatment can be administered to lighten the pains. To
perform this efficiently at the right time is complicated and
needs much knowledge about the disease and history of the
patient. It is essential to analyze the disease at the right time
and report its conditions. As hepatitis is a liver infection
disease, it may cause death if not diagnosed at the right
time. These are various symptoms for an abnormal liver.
The cause of hepatitis B includes the use of addictive drugs,
continuous use of alcohol and medicines, smoking, sharing
of daily use utensils with an infected person, blood
transfusion, sexual contact with infected person, etc. It is
common in areas where the system of sanitation is absent
and blood transfusion without proper protection is being
performed [2]. Many approaches for analysis have been
explored. Some of those are crucial physical examination,
liver tests, ultrasound, liver biopsy, blood tests, etc. Dif-
ferent blood tests are conducted for hepatitis B. After the
test of ALT [13] and AST, the major test is hepatitis B
surface antigen (HBsAg) [12, 18]. If the HBsAg test result is
positive, then other tests such as anti-HBsAg, anti-HBcAg,
anti-HBcAg-IgM, HBeAg, anti-HBeAg, and HBV-DNA
[17] must be conducted to check the level of hepatitis. If
chronic hepatitis is severe, it causes health issues. It can be
classified into five phases: (i) HBeAg-positive chronic in-
fection, (ii) HBeAg-positive chronic hepatitis, (iii) HBeAg-
negative chronic infection, (iv) HBeAg-negative chronic
hepatitis, and (v) HBsAg-negative phase [13]. Hepatitis-B
virus (HBV) infection is still a problem for global public
Hindawi
Journal of Healthcare Engineering
Volume 2019,Article ID 6361318, 11 pages
https://doi.org/10.1155/2019/6361318

health with substantial morbidity and mortality [13–16]. If
HBsAg is negative, then there are very fewer chances of
HBV. Sometimes HBsAg is negative and anti-HBsAg
(HBsAb) values are more than the cutoff values due to
some previous vaccination. This results in no-hepatitis B
state. In anti-HBcAg, anti-HBsAg is positive with negative
HBsAg which is due to the previous recovered HBV attack.
For acute hepatitis B, the HBsAg and anti-HBcAg-IgM
must be positive. If the test results of HBsAg and anti-
HBcAg are positive and anti-HBsAg and anti-HBcAg-IgM
are negative, it results in chronic hepatitis B. The proposed
ADHB-ML-MFIS expert system is based on these test
results. There are different data analysis techniques, and
some of them are based on machine learning, statistics, data
abstraction, decision support system, and expert system
[3]. Expert system techniques have been used in last few
years in medical analysis. They increase the diagnostic
accuracy and decrease the costs [4].
In all over the world, last-stage liver infection is a major
source of morbidity and death [17]. In 2015, according to the
World Health Organization (WHO), 1.34 million deaths
were occurred due to hepatitis and 257 million people were
infected with HBV worldwide [18]. In 2016, the WHO re-
ported that approximately 240 million people had chronic
hepatitis B virus infection from all over the world [19].
At present, artificial intelligence is being used to di-
agnose different kinds of medical problems. Intelligent
systems are being developed to resolve the medicals issues
[5]. Fuzzy inference system is the very powerful expert
system to analyze the problems and provide their solutions.
FIS is very useful where chances of uncertainty may occur. It
is used in every filed of life such as automatic robotics,
industries, computer sciences, medical systems, weather
forecasting, agriculture, and so on. Neshat et al. presented a
fuzzy system for the analysis and diagnosis of liver disorders
[4]. Obot and Udoh diagnosed hepatitis using the fuzzy
inference system on the basis of symptoms such as vomiting,
body weakness, nausea, bile in urine, loss of appetite,
jaundice, etc. [6]. Lancaster introduced a medical device on
the basis of fuzzy logic control (FLC). FLC is used for
managing the controller that employs air stress of human
skin, and to manage it, alarm was used [7]. Rana and
Sedamkar designed an expert system for medical diagnosis
using the fuzzy logic inference system [8]. Adeli et al. dis-
cussed and diagnosed hepatitis in their research. They in-
troduced “New Hybrid Hepatitis Diagnosis System Based on
Genetic Algorithm and Adaptive Network Fuzzy Inference
System” [9]. Dagar et al. introduced a FIS to diagnose
various diseases based on initial symptoms [10]. Umoh and
Ntekop proposed an expert system using the FIS to diagnose
and monitor cholera [11].
2. Methods
Our proposed automated diagnosis hepatitis B (ADHB)
multilayered mamdani fuzzy inference system- (MFIS-)
based expert system (ADHB-ML-MFIS ES) is explained in
this section. Figure 1 shows the flow of the proposed ADHB-
ML-MFIS expert system methodology.
The ADHB-ML-MFIS expert system consists of two
layers as shown in Figure 2. In layer I, hepatitis is diagnosed
(No/Yes) using two input variables, alanine aminotrans-
ferase (ALT) and aspartate aminotransferase (AST), as
shown in Figure 2.
The value of ALT and AST are also used to build up a
lookup table given in Table 1 to evaluate the status of
hepatitis. If layer I diagnoses hepatitis, then layer II is active.
Layer II diagnoses the stage of HB based on the seven input
variables as shown in Figure 2. Layer II input variables are
shown in Table 2.
The layer I of the proposed ADHB-ML-MFIS expert
system can be mathematically written as
μDH,layer1 � MFIS μALT, μAST􏼂 􏼃, (1)
and the layer II of the proposed ADHB-ML-MFIS expert
system can be expressed as
μDHB, layer2
� MFIS
μDH,layer1, μHBsAg, μanti−HBsAg , μanti−HBcAg,
μanti−HBcAg−lgM, μHBeAg, μanti−HBeAg, μHBV−DNA
⎡⎣ ⎤⎦.
(2)
2.1. Input Variables. Fuzzy input variables are statistical
values that are used to diagnose hepatitis B. In this search, a
total of nine different types of input variables are used on
both layers. Two variables are used at layer I, and rest of the
variables are used at layer II. The details of these input
variables with their ranges are shown in Tables 1 and 2.
Fuzzy crisp
inputs
Studying factors
that determine
Discard
Is the
factor
relevant?
Diagnose HB using
crisp outputs
No
Yes
Rules
Fuzzifier Defuzzifier
Inference
input sets
Fuzzy
output sets
Fuzzy
Crisp
outputs
Crisp
inputs
Figure 1: Proposed ADHB-MFIS-based expert system methodology.
2 Journal of Healthcare Engineering

2.2. Output Variables. In this search, multilayered archi-
tecture is proposed to diagnose hepatitis B. If the layer I
output is yes, then layer II is activated. Output variables for
both layers are shown in Table 3.
2.3. Membership Functions. The membership function of
this system gives curve values between 0 and 1 and also
provides a mathematical function that offers statistical
values of input and output variables. Graphical and math-
ematical representations of the proposed ADHB-ML-MFIS
expert system member functions of I/O variables of both
layers are shown in Table 4. These MFs are developed after
discussion with medical experts from Pathology De-
partment, Shalamar Hospital, Lahore, Pakistan.
2.4. Lookup Table. The lookup table for the proposed
ADHB-ML-MFIS-based expert system contains 50 input-
output rules. A few of them are shown in Table 5. This
lookup table is developed with the help of medical experts
from Pathology Department of Shalamar Hospital, Lahore,
Pakistan.
2.5.I/ORules. They play a critical role in any fuzzy inference
system (FIS). The performance of any expert system depends
upon these rules. In this research, I/O rules are developed
using a lookup table as shown in Table 6. Proposed I/O rule
based on the ADHB-ML-MFIS expert system is shown in
Figures 3 and 4.
2.6. Inference Engine. Inference engine is one of the core
components of any expert system. In this research, Mamdani
inference engine is used in both layers.
2.7.Defuzzifier. Defuzzifier is one of the critical components
of an expert system. There are different types of defuzzifiers.
In this research, a centroid-type defuzzifier is used. Figure 5
shows the defuzzifier graphical representation of layer I in
the ADHB-ML-MFIS expert system. In Figures 6(a)–6(d),
the graphical representations of the defuzzifier at the layer II
ADHB-ML-MFIS expert system is presented.
Table 1: Layer I input variables of the proposed ADHB-ML-MFIS
expert system.
Sr
no.
Input
parameters
Ranges Semantic sign
Reference
range/cutoff
value
1 AST
B/W
5–45 U/L
Normal
5–40 U/LB/W
40–550 U/L
Elevated values
GT > 500 Marked elevations
3 ALT
B/W
7–55 U/L
Normal
7–55 U/L
LT < 500 Elevated values
GT > 500 Marked elevations
LT � less than; GT �greater than; B/W � between; U/L � unit per liter.
Table 2: Layer II input variables of the proposed ADHB-ML-MFIS
expert system [13].
Sr
no.
Input
parameters
Ranges
Semantic
sign
Reference
range/cutoff
value
1 HBsAg
LT < 0.9 Negative
1.0B/W 0.9–1.0 Borderline
GT > 1.0 Positive
2 Anti-HBsAg
2–10 IU/L Negative
10 IU/L
GT > 10 Positive
3 Anti-HBcAg
LT < 1.0 Positive
GT > 1.0 Negative
4 Anti-HBcAg-IgM
LT < 1.0 Negative
GT > 1.0 Positive
5 HBeAg
LT < 0.67 Negative
0.67
GT > 0.67 Positive
6 Anti-HBeAg
LT < 0.75 Positive
0.75
GT > 0.75 Negative
7 HBV-DNA
LT < 10 Negative
10 IU/L
GT > 10 Positive
LT � less than; GT �greater than; B/W � between; IU/L � international unit
per liter; anti-HBsAg � HBsAb; anti-HBcAg-IgM � HBcAb-IgM; anti-
HBcAg � HBcAb; anti-HBeAg � anti-HBeAg.
Table 3: Layers I and II output variables of the proposed ML-
MFIS-DHB expert system.
Sr no.
Output
variables
Semantic sign
1 Layer I Hepatitis
No
Yes
Liver infection
2 Layer II DHB
No hepatitis B
Acute hepatitis
Chronic hepatitis
Immunity due to vaccination
Immunity due to the previous
infection
Layer I
Layer II
ALT
AST
Hepatitis
Yes/no
HBsAg
Anti-HBsAg
Anti-HBcAg
Anti-HBcAg-IgM
HBeAg
Anti-HBeAg
HBV-DNA
DHB
Figure 2: Proposed ADHB-ML-MFIS expert system.
Journal of Healthcare Engineering 3

Table 4: Input and output variables membership functions used in the proposed ADHB-ML-MFIS expert system.
Sr no. Input variables Membership function (MF) Graphical representation of MF
1 HBsAg � S (μS(s))
μS,N(s) � max(min(1, 0.9 − s/0.1), 0){ }
μS,BL(s) � max(min(s − 0.8/0.1, 1, 1.1 − s/0.1), 0){ }
μS,P (s) � max(min(s − 1/0.1, 1 ), 0){ }
0
Negative Borderline Positive
0.40.2 0.6 0.8
Input variable “HBsAg”
1 1.2 1.4 1.6 1.8 2
0
0.5
1
2
Anti-HBsAg � A
(μA (a))
μA,N(a) � max(min(1, 10.5 − a/0.1 ), 0){ }
μA,P(a) � max(min(a − 9.5/0.1, 1 ), 0){ }
Negative Positive
Input variable “Anti-HBsAg”
0
0
0.5
1
42 6 8 10 12 14 16 18 20
3
Anti-HBcAg � C
(μC(c))
μC,P(c) � max(min(1, 0.95 − c/0.05), 0){ }
μC,BL(c) � max(min(c − 0.9/0.05, 1, 1.1 − c/0.05), 0){ }
μC,N(c) � max(min(c − 1.05/0.05, 1 ), 0){ }
Positive Borderline Negative
0 0.40.2 0.6 0.8
Input variable “Anti-HBcAg”
1 1.2 1.4 1.6 1.8 2
0
0.5
1
4
Anti-HBcAg-IgM � I
(μI(i))
μI,N(i) � max(min(1, 0.95 − i/0.05), 0){ }
μI,BL(i) � max(min(i − 0.9/0.05, 1, 1.1 − i/0.05), 0){ }
μI,N(i) � max(min(i − 1.05/0.05, 1 ), 0){ }
Negative Borderline Positive
0
0
0.5
1
0.40.2 0.6 0.8
Input variable “Anti-HBcAg-lgM”
1 1.2 1.4 1.6 1.8 2

In Figure 5, diagnoses of hepatitis using probability are
based on two input parameters ALT and AST. If the values
of ALT and AST are elevated and ALT level is higher than
the AST level, then there is 80% chance for hepatitis to
occur. In this case, more than 80 % chances of hepatitis are
present. Our system diagnoses hepatitis. It is also observed
that if the AST level is higher than the ALT level, then there
is fair chance for hepatitis to occur. If both values of ALT
Table 4: Continued.
Sr no. Input variables Membership function (MF) Graphical representation of MF
5 HBeAg � E (μE(e))
μE,N(e) � max(min(1, 0.69 − e/0.04), 0){ }
μE,P(e) � max(min(e − 0.65/0.04, 1 ), 0){ }
Negative Positive
0
0.5
1
0 0.40.2 0.6 0.8
Input variable “HBeAg”
1 1.2
6
Anti-HBeAg � T
(μT(t))
μT,P(t) � max(min(1, 0.8 − t/0.1), 0){ }
μT,N(t) � max(min(t − 0.8/0.1, 1 ), 0){ }
NegativePositive
0
0.5
1
Input variable “Anti-HBeAg”
0 0.5 1 1.5
7
HBV-DNA � V
(μV(v))
μV,N(v) � max(min(1, 10.5 − v/0.1), 0){ }
μV,P(v) � max(min(v − 9.5/0.1, 1 ), 0){ }
Input variable “HBV-DNA”
0 42 6 8 10 12 14 16 18 20
0
0.5
1
Negative Positive
8
Hepatitis � H
(μH(h))
μH,N(h) � 0.25 − h/0.2, 0 ≤ h ≤ 0.25􏼈 􏼉
μH,A(h) �
h/0.25, 0 ≤ h ≤ 0.25
0.5 − h/0.25, 0.25 ≤ h ≤ 0.5
􏼨 􏼩
μH,C(h) �
h − 0.25/0.25, 0.25 ≤ h ≤ 0.5
0.75 − h/0.25, 0.5 ≤ h ≤ 0.75
􏼨 􏼩
μH,V(h) �
h − 0.5/0.25, 0.5 ≤ h ≤ 0.75
1 − h/0.25, 0.75 ≤ h ≤ 1
􏼨 􏼩
μH,I(h) � h − 0.75/0.25, 0.75 ≤ h ≤ 1􏼈 􏼉
Output variable “Hepatitis-condition”
0 0.20.1
No hepatitis B
Acute hepatitis
Chronic hepatitis
Due to vaccination
Due to infection
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1

Table 5: Lookup table for the proposed ADHB-ML-MFIS.
Rules HBsAg Anti-HBsAg Anti-HBcAg Anti-HBcAg (IgM) HBeAg Anti-HBeAg HBV-DNA Results
1 N N N — — — — None
2 N P N — — — — Due to vaccination
3 N P P — — — — Due to infection
4 P N P P — — —
Acute HBV
5 P N P P — P P
6 P N P P P — P
7 P N N P P N P
8 P N P P P N N
9 P N N P P P N
10 P N P N — — —
Chronic HBV
11 P N P N — — P
12 P N P N P — —
13 P N P N P — —
14 P N P N — P P
15 P N P N P — P
Table 6: Accuracy of the proposed ADHB-ML-MFIS expert system.
Patient
HBs
Ag
Anti-
HBsAg
Anti-
HBcAg
Anti-
HBcAg
(IgM)
HBe
Ag
Anti-
HBeAg
HBV-
DNA
Human
expert
decision
Proposed DHB-
ML-MFIS expert
system decision
Probability
of
correctness
Probability
of errors
1 N (0.2) N(7) N(1.9) — — — —
None None 1.0 0
2 N (0.4) N(5) BL(0.98) — — — —
3 N (0.5) P(20) N(1.5) — — — — Due to
vaccination
Due to vaccination 1.0 0
4 N (0.7) P(20) BL(0.97) — — — —
5 N (0.4) P(20) P(0.2) — — — —
Due to
infection
Due to infection
0.75 0.25
6 BL (0.91) P(20) P(0.55) — — — —
7 BL(0.91) P(20) P(0.37) — — — —
8 BL(0.95) P(20) P(0.31) — — — — Chronic HBV
9 P (1.7) N(3) P(0.44) P(1.67) — — —
Acute HBV
Acute HBV
0.95 0.05
10 P(1.9) N(5) BL(0.98) P(1.17) — — — Acute HBV
11 BL(0.95) N(7) P(0.75) P(1.31) — — — Acute HBV
12 BL(0.95) N(5) BL(0.99) P(1.57) — — — Acute HBV
13 P(1.7) N(6) P(0.73) P(1.43) P(1.11) — — Acute HBV
14 P(1.48) N(7) P(0.48) P(1.63) — P(0.56) P(17) Acute HBV
15 P(1.45) N(6) P(0.25) P(1.28) P(0.87) — P(20) Acute HBV
16 P(1.65) N(2) P(0.45) P(1.63) — N(1.23) P(20) Acute HBV
17 P(1.4) N(6) P(0.7) P(1.68) P(0.9) N(0.87) — Acute HBV
18 P(1.31) N(3) P(0.55) BL(0.99) P(0.77) P(0.45) — Acute HBV
19 P(1.42) N(6) P(0.37) BL(0.901) P(0.8) N(1.19) — Chronic HBV
20 P(1.3) N(4) P(0.5) BL(1.03) P(0.85) N(1.17) P(20) Acute HBV
21 P(1.57) N(3) P(0.35) BL(0.97) P(0.97) P(0.27) P(20) Acute HBV
22 BL(0.91) N(5) P(0.47) P(1.43) P(0.80) P(0.31) P(20) Acute HBV
23 P(1.9) N(4) P(0.75) P(1.29) P(0.99) P(0.65) P(20) Acute HBV
24 BL(0.93) N(8) P(0.21) P(1.65) P(0.89) N(1.05) P(20) Acute HBV
25 P(1.8) N(6) BL(1.01) P(1.43) P(0.93) N(1.17) P(20) Acute HBV
26 P(1.31) N(5) N(1.7) P(1.3) P(0.87) N(1.0) P(20) Acute HBV
27 P(1.7) N(8) P(0.72) P(1.29) P(0.97) N(0.87) N(7) Acute HBV
28 P(1.4) N(3) N(1.38) P(1.57) P(0.73) P(0.35) N(5) Acute HBV
29 P(1.21) N(6) P(0.51) P(1.81) N(.35) P(0.49) N(7) Acute HBV

and AST are in the normal range, then it means no
hepatitis.
Figure 6(a) shows hepatitis B (regarding probability) based
on HBsAg and anti-HBsAg. Different colours in the surface
region present the stages of hepatitis. It is also observed that if
anti-HBsAg (x-axis) is negative (equivalent mathematically
lies between 2 and 10 IU/L) and HBsAg (y-axis) is less than
0.8, then the probability of hepatitis B (z-axis) is 0; that is, it
may be any other type of hepatitis. It is also observed that if
costs of anti-HBsAg is more the 10 IU/L its mean positive,
amounts of HBsAg is less the 0.8, and the value of hepatitis is
80% which is due to vaccination or some previous infection.
Similarly, remaining Figures 6(b)–6(d) present hepa-
titis B results by prevailing different input parameter
values. The surface region represents probability values by
two input variables from the given seven input variables.
The hepatitis B results are the combination of at least three
input variables.
3. Results
For simulation results, MATLAB R2017a tool is used.
MATLAB is also used for modelling, simulation, algorithm
development, prototyping, and many other fields. MATLAB
Table 6: Continued.
Patient
HBs
Ag
Anti-
HBsAg
Anti-
HBcAg
Anti-
HBcAg
(IgM)
HBe
Ag
Anti-
HBeAg
HBV-
DNA
Human
expert
decision
Proposed DHB-
ML-MFIS expert
system decision
Probability
of
correctness
Probability
of errors
30 P(1.42) N(5) P(0.37) N(0.41) — — —
Chronic
HBV
Chronic HBV
0.91 0.09
31 P(1.71) N(7) BL(0.93) N(0.49) — — — Chronic HBV
32 P(1.48) N(2) BL(1.08) N(0.68) — — P(20) Chronic HBV
33 P(1.2) N(4) P(0.2) N(0.2) — — P(20) Chronic HBV
34 P(1.7) N(3) P(0.25) N(0.47) P(1.2) — — Chronic HBV
35 P(1.3) N(8) P(0.65) N(0.19) P(0.92) — — Chronic HBV
36 P(1.5) N(7) P(0.72) N(0.23) — P(0.37) P(20) Chronic HBV
37 P(1.21) N(3) P(0.23) N(0.51) P(0.89) — P(20) Chronic HBV
38 P(1.35) N(5) BL(1.02) N(0.45) P(0.99) — P(20) Chronic HBV
39 P(1.5) N(9) P(0.72) N(0.39) P(1.0) N(1.28) — Chronic HBV
40 P(1.9) N(4) P(0.15) N(0.23) — N(0.92) P(20) Chronic HBV
41 P(1.4) N(6) BL(0.93) N(0.76) — P(0.48) P(20) Chronic HBV
42 BL(0.91) N(7) P(0.63) N(0.45) — P(0.93) P(20) Chronic HBV
43 BL(0.902) N(3) P(0.27) N(0.23) P(0.92) — P(20) Chronic HBV
44 BL(0.902) N(5) BL(1.02) N(0.39) P(0.82) — P(20) Acute HBV
45 P(1.2) N(6) BL(0.93) N(0.18) P(0.95) N(1.4) P(20) Chronic HBV
46 P(1.4) N(6) P(0.3) N(0.47) P(0.89) N(1.15) P(20) Chronic HBV
47 P(1.7) N(8) BL(0.93) N(0.71) N(0.45) P(0.45) N(5) Chronic HBV
48 P(1.3) N(3) P(0.85) N(0.65) N(0.32) P(0.38) N(4) Chronic HBV
49 P(1.7) N(7) BL(0.93) N(0.42) N(0.47) P(0.52) P(20) Chronic HBV
50 BL(0.93) N(7) P(0.86) N(0.28) N(0.31) P(0.35) P(20) No hepatitis
51 BL(.91) N(4) BL(0.96) N(0.67) N(0.37) P(0.37) P(20) Chronic HBV
52 P(1.3) N(5) P(0.25) N(0.47) N(0.46) P(61) P(20) Chronic HBV
Figure 3: Layer I I/O rules for the proposed ADHB-ML-MFIS expert system.

is an efficient tool for programming, data analysis, visuali-
zation, and computing. For simulation results, nine inputs
and one output DHB variables are used. When results of
layer I show hepatitis, there can be different types of hepatitis
such as hepatitis A, B, C, D, and E. In this research, the
proposed ADHB-ML-MFIS-based expert system not only
diagnosed hepatitis B but also showed the different levels of
hepatitis B such as acute, chronic, etc. But if layer I diagnoses
hepatitis and layer II diagnoses no hepatitis B, its means that
it may contain other types of hepatitis. Figures 7(a)–7(c)
show the performance of the proposed ADHB-ML-MFIS
expert system at layer I.
Figure 7(a) shows that if the values ALT and AST are in
the normal range, then there is no hepatitis or other in-
fections. Figure 7(b) shows that if the values of AST are
greater than ALT, then the elevation may be due to alcohol
or any other problem. Figure 7(c) shows the high cost of
ALT as it is more elevated than AST showing hepatitis.
Table 6 shows the accuracy of the proposed ADHB-ML-
MFIS expert system in comparison with Medical Human
expert of Shalamar Hospital, Lahore, Pakistan. The efficiency
of the proposed method is randomly checked on 52 records.
The standard unit of anti-HBsAg and HBV-DNA is IU/L;
during simulation in most cases, we considered their values
are 20 IU/L. The proposed DHB-ML-MFIS expert system
provides the accurate results for all costs, and only at
borderline it may achieve some minor errors.
Figure 8 shows the precision of the proposed ADHB-
ML-MFIS expert system in the form of probability of all
output cases. The last column produces an overall efficiency
of the proposed ADHB-ML-MFIS expert system which is
92.2%.
4. Conclusion and Future Work
The primary focus of our research was to design an expert
system to diagnose hepatitis B by ELISA blood test reports
taken from Pathology Department of Shalamar Hospital,
Lahore, Pakistan. The proposed expert system is elementary
and easy to use for both medical and nonmedical pro-
fessionals. A common man can also diagnose the status of
hepatitis by providing required inputs. The primary objective
of this research is to diagnose the different levels of hepatitis B.
The overall precision of the proposed DHB-ML-MFIS expert
550
600
500450400350300
200 250
15010050
0.5
0.6
0.7
0.8
0.4
0.3
0.2
AST (U/L)
ALT (U/L)
600
Results
500
550
450
400
350
300
250
200
150 100
50
Figure 5: Layer I rules surface for ALT and AST.
Figure 4: Layer II I/O rules for the proposed DHB-ML-MFIS expert system.

1
ALT (U/L) = 31.4 AST (U/L) = 26 Results = 0.2
2
3
4
5
6
7
0 0600 600
0 1
(a)
ALT (U/L) = 178 AST (U/L) = 347 Results = 0.5
1
2
3
4
5
6
7
0 600 0 600
0 1
(b)
Figure 7: Continued.
Hepatitiscondition
21.81.61.41.21
0.6 0.8
0.40.20 HBsAg
Anti-HBsAg
0.7
0.6
0.5
0.4
0.3
0.2
0.1
20 18
16 14 12 10 8 6 4 2
0
(a)
Hepatitiscondition
21.81.61.41.21
0.6 0.8
0.40.20 HBsAg
Anti-HBcAg
0.8
0.85
0.9
0.75
0.7
0.65
0.6
0.55
0.5
2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
(b)
Hepatitiscondition
201816141210
6 8
420
0.75
0.7
0.65
0.6
0.55
0.5
2 1.8 1.6 1.4 1.2 1 0.8
0.6 0.4 0.2 0
Anti-HBcAg Anti-HBsAg
(c)
Hepatitiscondition
201816141210
6 8
420
0.6
0.7
0.5
0.4
0.3
2 1.8 1.6 1.4 1.2 1 0.8
0.6 0.4 0.2 0
Anti-HBcAg-IgM Anti-HBsAg
(d)
Figure 6: Rule surface for (a) anti-HBsAg and HBsAg, (b) anti-HBcAg and HBsAg, (c) anti-HBcAg and anti-HBsAg, and (d) anti-HBcAg-
LGM and anti-HBsAg.

system is 92.2%. In future, the efficiency of the proposed
system can be improved using other techniques including
computational intelligence such as neural network and
neurofuzzy systems. This research work can be extended to
others types of hepatitis such as A, C, D, and E.
Data Availability
The clinical/patient data used to support the findings of this
study are restricted by the Shalamar Hospital, Lahore,
Pakistan, in order to maintain patient privacy. The simu-
lation files/data used to support the findings of this study are
available from the corresponding author upon request.
Conflicts of Interest
The authors declare that there are no conflicts of interest
regarding the publication of this article.
Acknowledgments
The authors would like to express their deepest gratitude to
Mrs. Niala Naz from UOL for the helpful suggestions during
data collection and result interpretation.
References
[1] J. M. Ntaganda and M. Gahamanyi, “Fuzzy logic approach for
solving an optimal control problem of an uninfected hepatitis
B virus dynamics,” Applied Mathematics, vol. 6, no. 9,
pp. 1524–1537, 2015.
[2] P. A. Ejegwa and E. S. Modom, “Diagnosis of viral hepatitis
using new distance measure of intuitionistic fuzzy sets,” In-
ternational Journal of Fuzzy Mathematical Archive, vol. 8,
no. 1, pp. 1–7, 2015.
[3] N. Cheung, “Machine learning techniques for medical anal-
ysis,” Doctoral dissertation, B. Sc. Thesis, School of In-
formation Technology and Electrical Engineering, University
of Queenland, Brisbane, Australia, 2001.
[4] M. Neshat and M. Yaghobi, “Designing a fuzzy expert system
of diagnosing the hepatitis B intensity rate and comparing it
with adaptive neural network fuzzy system,” in Proceedings of
World Congress on Engineering and Computer Science,
pp. 797–802, San Francisco, USA, October 2009.
[5] A. Sardesai, P. Sambarey, V. Kharat, and A. Deshpande,
“Fuzzy logic application in gynecology: a case study,” in
Proceedings of 2014 International Conference on Informatics,
Electronics and Vision (ICIEV), pp. 1–5, IEEE, Dhaka, Ban-
gladesh, May 2014.
ALT (U/L) = 359 AST (U/L) = 176 Results = 0.8
1
2
3
4
5
6
7
0 600 0 600
(c)
Figure 7: Layer I. Lookup diagram for (a) normal, (b) the other liver infections, and (c) hepatitis.
1 1
0.75
0.95 0.91 0.922
0 0
0.25
0.05 0.09 0.078
No hepattis Due to
vaccination
Due to
infection
Acute HBV Chronic HBV Overall
results
Precision chart
Correctness prob.
Error’s prob.
Figure 8: Precision of the proposed ADHB-ML-MFIS expert system.

[6] O. U. Obot and S. S. Udoh, “A framework for fuzzy diagnosis
of hepatitis,” in Proceedings of 2011 World Congress on In-
formation and Communication Technologies (WICT),
pp. 439–443, IEEE, Mumbai, India, December 2011.
[7] S. S. Lancaster, “A fuzzy logic controller for the application of
skin pressure,” in Proceedings of 2004 IEEE Annual Meeting of
the Fuzzy Information Processing NAFIPS’04, pp. 686–689,
IEEE, Alberta, Canada, 2004.
[8] M. Rana and R. R. Sedamkar, “Design of expert system for
medical diagnosis using fuzzy logic,” International Journal of
Scientiﬁc and Engineering Research, vol. 4, no. 6, pp. 2914–
2921, 2013.
[9] M. Adeli, N. Bigdeli, and K. Afshar, “New hybrid hepatitis
diagnosis system based on Genetic algorithm and adaptive
network fuzzy inference system,” in Proceedings of 2013 21st
Iranian Conference on Electrical Engineering (ICEE), pp. 1–6,
IEEE.
[10] P. Dagar, A. Jatain, and D. Gaur, “Medical diagnosis system
using fuzzy logic toolbox,” in Proceedings of 2015 In-
ternational Conference on Computing, Communication and
Automation (ICCCA), pp. 193–197, IEEE, Greater Noida,
India, May 2015.
[11] U. A. Umoh and M. M. Ntekop, “A proposed fuzzy framework
for cholera diagnosis and monitoring,” International Journal
of Computer Applications, vol. 82, no. 17, 2013.
[12] M. Cornberg, V. W.-S. Wong, S. Locarnini, M. Brunetto,
H. L. A. Janssen, and H. L.-Y. Chan, “The role of quantitative
hepatitis B surface antigen revisited,” Journal of hepatology,
vol. 66, no. 2, pp. 398–411, 2017.
[13] European Association for the Study of the Liver, “EASL 2017
Clinical Practice Guidelines on the management of hepatitis B
virus infection,” Journal of hepatology, vol. 67, no. 2,
pp. 370–398, 2017.
[14] A. Craxi, H. Wedemeyer, K. Bjoro et al., “European associ-
ation for the study of the liver. EASL clinical practice
guidelines: management of hepatitis C virus infection,”
Journal of hepatology, vol. 55, 2011.
[15] A. Schweitzer, J. Horn, R. T. Mikolajczyk, G. Krause, and
J. J. Ott, “Estimations of worldwide prevalence of chronic
hepatitis B virus infection: a systematic review of data pub-
lished between 1965 and 2013,” The Lancet, vol. 386,
no. 10003, pp. 1546–1555, 2015.
[16] R. Lozano, M. Naghavi, K. Foreman et al., “Global and re-
gional mortality from 235 causes of death for 20 age groups in
1990 and 2010: a systematic analysis for the Global Burden of
Disease Study 2010,” The lancet, vol. 380, no. 9859,
pp. 2095–2128, 2012.
[17] J. F. Perz, G. L. Armstrong, L. A. Farrington, Y. J. F. Hutin, and
B. P. Bell, “The contributions of hepatitis B virus and hepatitis C
virus infections to cirrhosis and primary liver cancer world-
wide,” Journal of hepatology, vol. 45, no. 4, pp. 529–538, 2006.
[18] World Health Organization, Global Hepatitis Report 2017,
World Health Organization, Geneva, Switzerland, 2017.
[19] World Health Organization, Global Health Sector Strategy on
Viral Hepatitis 2016-2021. Towards Ending Viral Hepatitis,
World Health Organization, Geneva, Switzerland, 2016.

International Journal of
Aerospace
EngineeringHindawi
www.hindawi.com Volume 2018
Robotics
Journal of
Hindawi
Hindawi
Active and Passive
Electronic Components
VLSI Design
Hindawi
Hindawi
Shock and Vibration
Hindawi
Civil Engineering
Advances in
Acoustics and Vibration
Advances in
Hindawi
Hindawi
Electrical and Computer
Engineering
Journal of
Advances in
OptoElectronics
Hindawi
www.hindawi.com
Volume 2018
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2013
Hindawi
www.hindawi.com
The Scientific
World Journal
Volume 2018
Control Science
and Engineering
Journal of
Hindawi
Hindawi
www.hindawi.com
Journal of
Engineering
Volume 2018
Sensors
Journal of
Hindawi
Rotating
Machinery
Hindawi
Modelling &
Simulation
in Engineering
Hindawi
Hindawi
Chemical Engineering
International Journal of Antennas and
Propagation
Hindawi
Hindawi
Navigation and
Observation
Hindawi
Advances in
Multimedia
Submit your manuscripts at
www.hindawi.com

Automated diagnosis of hepatitis b using multilayer mamdani

More Related Content

What's hot (19)

Similar to Automated diagnosis of hepatitis b using multilayer mamdani (20)

Recently uploaded (20)

Automated diagnosis of hepatitis b using multilayer mamdani