A fuzzy logic based expert system for determination of health risk level of patient

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 04 Issue: 05 | May-2015, Available @ http://www.ijret.org 261
A FUZZY LOGIC BASED EXPERT SYSTEM FOR DETERMINATION
OF HEALTH RISK LEVEL OF PATIENT
Monish Kumar Choudhury1
, Neelanjana Baruah2
1
Student M.E. Electrical Engineering, Jorhat Engineering College, Assam, India
2
Department of Electrical Engineering, Jorhat Engineering College, Assam, India
Abstract
The aim of this study is to design a fuzzy expert system for calculating the health risk level of a patient. The fuzzy logic system is a
simple, rule-based system and can be used to monitor biological systems that would be difficult or impossible to model with
simple, linear mathematics. The designed system is based on the modified early warning score (MEWS).The system has 5 input
field and 1 output field. The input fields are blood pressure, pulse rate, SPO2 ( it is an estimation of the oxygen saturation level in
blood. ), temperature, and blood sugar. The output field refers the risk level of the patient. The output ranges from 0 to 14. This
system uses Mamdani inference method. A larger value of output refers to greater degree of illness of the patient. This paper
describes research results in the development of a fuzzy driven system to determine the risk levels of health for the patients. The
implementation and simulation of the system is done using MATLAB fuzzy tool box.
Keywords: Fuzzy logic, The Modified Early Warning Score (MEWS), Physiological Parameters, Classification of
Vital Signs, MATLAB Tool, Fuzzy Inference System, Fuzzification, Defuzzification
--------------------------------------------------------------------***----------------------------------------------------------------------
1. INTRODUCTION
In day to day life there are many situations in which it is
very useful if we can determine the risk level of patients.
For example occurrence of a natural digester like flood,
earthquake etc. affects people in many fields of their life
including their health. In such situations many people die
because common people around them cannot predict that
they require immediate treatment and also it became very
difficult for the doctors to reach them. So if there is a system
that can predict the health status of the people depending
upon their physiological parameters without the help of a
doctor then it will be very helpful for saving the life of many
people. Motivated by the need of such an important system,
in this study an expert system is designed to determine the
risk level of patient so as to predict their health status. Now
risk level calculation using the physiological parameters
involves lots of inaccuracy and uncertainty. In such situation
fuzzy logic can give much satisfactory result since it can
provide accurate information when there is inaccuracy
[1].The expert system is designed using Fuzzy Logic. In
this work, the fuzzy logic is based on the modified early
warning score (MEWS), which is a simple guide used by
hospital nursing and medical staff as well as emergency
medical services to quickly determine the degree of illness
of a patient [2]. This fuzzy control system has been
implemented in MATLAB Tool. In this paper introduced
fuzzy control system to design fuzzy rule base to analyze the
risk level of patient health and the rule viewed by surface
view.
2. THE FUZZY LOGIC SYSTEM
Fuzzy Logic provides an effective tool for describing the
characteristics of a system that is too complex or ill-defined
to admit precise mathematical analysis. This theory is based
on approximate reasoning which plays a major role in
human thought process. Fuzzy logic is an Artificial
Intelligence technique which has the ability to mimic human
mind in terms of approximate reasoning rather than being
exact [3]. A Fuzzy Set has values with partial membership
along with the crisp values. Fuzzy Sets are useful in
establishing conditions which are imprecise in definition
through partial membership values. Elements in fuzzy set
can overlap, so a given crisp value can belong to multiple
fuzzy sets with different membership degrees in each set[4].
To utilize fuzzy logic, four components are required:
fuzzification, an inference, a fuzzy rule base, and
defuzzification [5]. One of the basic principles of fuzzy
logic is the degree of membership determined by
“fuzzifying” each data point using the input fuzzy set. The
input fuzzy set is determined by the system designer to
break down the complete range of possible input values into
membership functions. Each membership function has a
value of either 0 or 1 and a minimum and maximum range
of input value. Several shapes for the membership function
can be used, including trapezoidal, Gaussian, and triangular.
The most common and simplest to understand are
trapezoidal and triangular shaped membership functions,
which can be assembled into a fuzzy set by setting the
minimum input value of each function to the center point of
the previous membership function.

_______________________________________________________________________________________
3. MODIFIED EARLY WARNING SCORE
(MEWS)
Early warning scoring tools are used to aid recognition of
deteriorating patients, and are based on physiological
parameters, which are taken when recording patient
observations. The observations incorporated in this scoring
system should include: temperature, pulse, blood pressure
and respiratory rate, with oxygen saturations, level of
consciousness and urine output. An aggregated score is then
calculated from all seven parameters. There is an identified
threshold score which, when reached, then activates an
escalation pathway. The escalation pathway outlines actions
required for timely review ensuring appropriate
interventions for patients. It should be remembered that
MEWS is an aid to good clinical judgment, not a substitute
for it.[6] It can be used to quickly identify patients who are
clinically failing and who need urgent intervention. MEWS
can be used to monitor medical patients during assessment
and transport. The use of MEWS has been shown to be
effective in reducing death rates and illness chances of
patients whose health slowly worsens. MEWS can be used
to monitor medical patients during assessment and transport.
The use of MEWS has been shown to be effective in
reducing death rates and illness chances of patient’s health.
Table 1: The Modified Early Warning Score [7]
MEWS +3 +2 +1 0 +1 +2 +3
Systolic blood
pressure
<70 70-80 81-100 101-199 ≥200
Heart rate <40 41-50 51-100 101-110 111-130 >130
Respiratory
rate
<9 9-14 15-20 21-29 ≥30
Temperature <35
AVPU/GCS
score
<9 9-13 14 A/15 v/confused P U
AVPU=Alert, Verbal, Pain, Unresponsive; GCS=Glasgow Coma Scale
A MEWS is calculated for a patient using the five simple
physiological parameters shown in Table 1. Respiratory rate,
heart rate, systolic blood pressure, temperature and AVPU.
A score is given to a specific range of values for each of the
parameters in the table. The patient’s data for each
parameter is cross referenced against the MEWS table and a
score from 0 to 3 is allocated. The score for each parameter
is then added to give the MEWS score. A score of zero
shows that the patient case is normal, a score that is more
than zero and less than five shows that the patient is in a
Low Risk case, and a score of five or more shows that the
patient is in a High Risk case, and an admission to an
intensive care unit is recommended. In this work, different
MEWS parameters were used in order to calculate the
MEWS score. The parameters used are: systolic blood
pressure (SBP), heart rate (HR), oxygen saturation (SPO2),
body temperature (TEMP), and blood sugar (BS) .An
expert’s knowledge was used for dividing the input fields.
4. FUZZY EXPERT SYSTEM DESIGNING
MATLAB fuzzy tool box is used for implementation and
simulation of the fuzzy expert system. In order to design a
fuzzy expert system the typical steps followed are
determination of the input and output variables, the selection
of suitable membership functions, and the creation of the
fuzzy rules database [8]. FIS editor used for defining the
input and output variables is shown in Figure.1. Again the
membership function editor used for defining the
membership functions of each input and output variable is
shown in Figure.2.
4.1 Input Variables
For designing the expert system five input variables i.e.
blood pressure, heart rate, SPO2, temperature and blood
sugar are used. These inputs are called vital signs and use to
predict the health status of person. After choosing the input
variables the next step is to fuzzify the variables i.e. we have
to determine the fuzzy sets for each input variable and the
corresponding range of the belonging to each fuzzy set.
Fig. 1: Mamdani FIS editor with 5 inputs & 1 output

_______________________________________________________________________________________
Fig. 2: Membership function editor
Depending on the MEWS scoring system and using the
expert advice fuzzy sets for each of input variables are
determined. Membership functions of the fuzzy sets are
taken as trapezoidal. For blood pressure we use only the
systolic value. To fuzzify the SBP variable, range of values
for SBP which would be considered as normal are needed.
Let this be 100 to 185mm Hg (not everyone might agree
with this, so this choice merely captures the experience of
one particular expert). Thus a fuzzy set labeled Normal-0 is
created and values of SBP between 101 and 199 mm Hg to a
membership level of 1.0 is assigned to this set. Next we
address the more vague issue of what range of values for
SBP could possibly be normal but also be abnormal. Per the
expert advice, the range 185 to 199 was decided to be at the
upper end and 95 to 100 at the first lower end. In other
words, if SBP is above 199 mm Hg it is unquestionably too
high (which is labeled High 2 in the Fuzzy set), whereas
between 185 and 199 mm Hg, it could go either way. Same
procedure is followed for all the other input variables for
determining the fuzzy sets and the membership function.
The fuzzy sets form by the classification of each vital sign
and the corresponding membership functions are shown
below.
Table 2: Classification of Systolic Blood Pressure
Input Field Range Fuzzy Sets
Systolic Blood
Pressure
<75 Low-3
70 – 85 Low-2
80 – 100 Low-1
95 – 199 Normal-0
>185 High-2
Fig.3: Membership function of systolic blood pressure
Table 3: Classification of Heart Rate
Heart Rate
<50 Low- 2
45 - 60 Low-1
53 -100 Normal
95 – 110 High -1
105- 130 High-2
>125 High -3
Fig.4: Membership function of Heart Rate
Table 4: Classification of SPO2
SPO2
<85 Low -3
83- 90 Low-2
87 -95 Low-1
>93 Normal -0

_______________________________________________________________________________________
Fig.5: Membership function of SPO2
Table 5: Classification of Temperature
Temperature <36.5 Low 2
36 – 38.5 Normal 0
>38 High2
Fig.6: Membership function of Temperature
Table 6: Classification of Blood Sugar
Blood Sugar
<66 Low -3
63 -72 Low-2
70 -110 Normal-0
106 -150 High-2
>140 High- 3
Fig.7: Membership function of blood Sugar
4.2. Output Variables
In this fuzzy expert system there is one output variable i.e.
the Risk Level, which refers to the degree of illness of the
patient. Larger the value of this output variable more will be
the health risk of the patient. In this system, we have 15
fuzzy sets for the output variable risk level (NRM, LRG1,
LRG2, LRG3, LRG4, HRG5, HRG6, HRG7, HRG8, HRG9,
HRG10, HRG11, HRG12, HRG13, and HRG14).
Membership functions for these sets are triangular.
Fig.8: Membership function of Output variable (Risk Level)
Table 7: Classification of Output variable (Risk Level)
Input
Field
Range Fuzzy Sets
0<RL<0.5 NRM
0.5<RL<1.5 LRG1
1.5<RL<2.5 LRG2
2.5<RL<3.5 LRG3
3.5<RL<4.5 LRG4
4.5<RL<5.5 HRG 5
5.5<RL<6.5 HRG 6
6.5<RL<7.5 HRG 7

_______________________________________________________________________________________
Risk
Level
7.5<RL<8.5 HRG 8
6.5<RL<9.5 HRG 9
6.5<RL<10.5 HRG 10
6.5<RL<11.5 HRG 11
6.5<RL<12.5 HRG 12
6.5<RL<13.5 HRG 13
13.5<RL<14 HRG 14
4.3. Fuzzy Rule Base
The rule base is the main part in the fuzzy inference system
and the quality of results in a fuzzy system depends on the
fuzzy rules. The designed expert system in this work
includes 1800 rules that cover all possible cases. The
numbers of rules were obtained using the formula of
Equation
N=I(1)×I(2)×I(3)×I(4)×I(5)×I(6)……×I(n)
Where N is the total number of possible rules for a fuzzy
system and I(n) is the number of linguistic terms for the
input linguistic variable n. The rules for this fuzzy expert
system were formulated using MEWS scoring system. As all
inputs are dependent on each other, therefore in this system
we use logical combination of inputs with AND because all
the inputs are dependent on each other. The results with the
1800 rules tend to be similar to the MEWS scoring system.
A sample of the rules has been shown in Figure 9.
Fig.9: Sample of the Fuzzy Logic System Rules
4.4 Fuzzification and Defuzzification
Fuzzification is the first step in the design of any fuzzy
expert system. It is the process of mapping a crisp value of
an input to membership degrees in different Fuzzy
Linguistic variables.. Defuzzification is the inverse process
of fuzzification. It is the process of combining fuzzy output
of all the rules to give one crisp value. Thus crisp value
output is given by the defuzzification process after
estimating its input value. An example of the designed
system results in MATLAB is shown in Figure 10.
The following values are given to each input field: Systolic
Blood Pressure (SBP) =120, Heart Rate (HR) =75,
SPO2=98, Temperature (TEMP) =37 and Blood Sugar (BS)
=95. The fuzzy logic engine is triggered. The MATLAB-
rule viewer and simulation results are shown in Figure10
.New input values generate new depression risk output
responses.
Fig.10: Rule viewer showing a simulation result of the
designed system

_______________________________________________________________________________________
Surface viewer of some fields as follow
Fig.11. Surface Viewer of Blood Sugar and Heart Rate
Fig.12. Surface Viewer of temperature and SPO2
Fig.13. Surface Viewer of SPO2 and Heart Rate
Fig.14. Surface Viewer of temperature and blood sugar
5. RESULT AND DISCUSSION
The designed expert system has been tested with some set of
values of patient’s vital signs as shown in Table 8.. Also a
comparison between the MEWS scoring system results and
fuzzy logic results has been done in order to evaluate the
performance of the designed system.
Now in Table 8, total 6 cases are considered. Case 1 and
Case 2 corresponds to normal conditions of the patient. For
these two cases score according to MEWS system is 0 while
the designed expert system gives a score of .126 for each
case which is also very close to 0 and hence the result is
quite satisfactory. Again Case 3 and Case 4 corresponds
conditions when patient is under low risk. The score
according to MEWS system for Case 3 is 2 and that for Case
4 is 3. For these two cases the designed expert system gives
a score of 2 and 3. Similarly Case 5 and Case 5 correspond
when the patient is under high risk. For these two cases also
the result of the designed system is satisfactory.

_______________________________________________________________________________________
Table 8: A comparison between the mews results and the fuzzy logic results
Sl No Vital Signs Result
Using MEWS Using Fuzzy Logic
SBP HR SPO2 TEMP BS Status Score Status Score
1 120 65 95 37 88 NRM 0 NRM .126
2 170 80 96 37.5 102 NRM 0 NRM .126
3 117 63 98 38.2 115 LRL 2 LRL 2.42
4 180 92 97 37 220 LRL 3 LRL 3
5 187 87 97 41 161 HRL 5 HRL 4.67
6 190 120 94 38.6 120 HRL 6 HRL 6.82
6. CONCLUSION
The paper describes a design of a fuzzy expert system for
determination of the risk level of patient, which can be used
in any situation when it is necessary to predict the health
status of patient. The designed system can be used by the
doctor or by the patient himself. In this paper it can be
concluded that using expert knowledge embedded as fuzzy
rules and supplied input data (i.e. patent’s vital signs) ,the
designed system predicts risk level of patient and it can be
easily verified by the comparison done in table 9. Finally it
can be concluded that up to some extent the designed system
can be used as a virtual doctor.
REFERENCES
[1]. M. Mayilvaganan, K. Rajeswari, “Risk Factor Analysis
to Patient Based on Fuzzy Logic Control System” ,
International Journal of Engineering Research and General
Science Volume 2, Issue 5, August-September, 2014 ISSN
2091 -2730
[2]. The Free Encyclopedia From
Wikipedia."Http://En.Wikipedia.Org/Wiki/Mews,"
[3]. Mansoor Mirza, Hamid GholamHosseini, Member,
IEEE, Michael J. Harrison, “A Fuzzy Logic-based System
for AnaesthesiaMonitoring”, 32nd Annual International
Conference of the IEEE EMBS Buenos Aires, Argentina,
August 31 - September 4, 2010
[4]. M. Mayilvaganan, K.Rajeswari, “Health Care Analysis
Based On Fuzzy Logic Control System”, International
Journal of Computer Science Trends and Technology
(IJCST) , Volume 2 Issue 4, Jul-Aug 2014
[5]. Smita Sushil Sikchi, Sushil Sikchi, Ali M. S, “Generic
Medical Fuzzy Expert System for Diagnosis of Cardiac
Diseases” , International Journal of Computer Applications
(0975 – 8887) Volume 66– No.13, March 2013
[6]. Adult Modified Early Warning Score (MEWS) Policy
and Escalation Pathway Version 3
[7]. MultiCare Health System: Using a Modified Early
Warning System (MEWS) to Improve Patient Safety
HIMSS InnovationCommunity November 2, 2012
[8]. Ali.Adeli, Mehdi.Neshat , “A Fuzzy Expert System for
Heart Disease Diagnosis”, Proceeding of international
Multiconference of Engineers and Computer Scientists
2010 Vol I,IMECS 2010,March 17-19,2010,Hong Kong
[9]. L.A. Zadeh,”fuzzy theory,” university of
California,Berkely, 1965
[10]. "Understanding Fuzzy Logic: An Interview With Lotfi
Zadeh," IEE Signal Processing Magazine, May 2007.
[11]. Paul Grant, "A New Approach To Diabetic Control:
Fuzzy Logic And Insulin Pump Technology," Department
Of Medicine, Royal Sussex County Hospitall,Medical
Engineering & Physics, 2007.
[12]. Smita Sushil Sikchi, Sushil Sikchi, Ali M. S. “Fuzzy
Expert Systems (FES) for Medical Diagnosis”,
International Journal of Computer Applications (0975 –
8887) Volume 63– No.11, February 2013
[13]. L. Zadeh, "Fuzzy Sets," Information And Control, Vol.
8, No. 3, 1965

A fuzzy logic based expert system for determination of health risk level of patient

More Related Content

What's hot (18)

Similar to A fuzzy logic based expert system for determination of health risk level of patient (20)

Recently uploaded (20)

A fuzzy logic based expert system for determination of health risk level of patient