sysmodelling shon assignment 4
TRANSCRIPT
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 1/17
AUCKLAND UNIVERSITY OF TECHNOLOGY
Fuzzy Logic Approach in
Health CareAssignment 4
Shon U Johnny
Student ID : 0947283
Master Of Engineering Studies
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 2/17
Abst t
This Assi
¡ ¢
£
¡
t deals with the appli¤
ati
¥ ¡
of Ar tif i¤
ial Intelli
ence in medicine and health caresegments. This summar i¦ es the var ious fuzzy approaches taken. It explains about the changing face of
modern health care, application of ar tif icial intelligence in health care. It descr i bes the implementation
of intelligent alarms in Cardio Anaesthesia using fuzzy logic (Becker, H Kasmacher, Kalff, &
Zimmerman, 1994) and implementation of an automatic monitor ing and estimation tool for
cardiovascular system under ventr icular assistance using fuzzy reasoning. (Yoshizawa, Takeda,
Yambe, & Nitta, 1992). It also descr i bes the fuzzy logic Applications for Automatic control,
Supervision and Fault diagnosis (Iserman, 1998) It also evaluates a special ar ticle on anaesthetic
mishaps. (Gaba, Maxwell, & De Anda, 1987)
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 3/17
Introduction
Fuzzy logic is a form of multi-valued logic der ived from fuzzy set theory to deal with
reasoning that is approximate rather than precise. In contrast with "cr isp logic", where binary
sets have binary logic, the fuzzy logic var iables may have a membershi p value of not only 0
or 1 ± that is, the degree of truth of a statement can range between 0 and 1 and is not
constrained to the two truth values of classic propositional logic. Fur thermore, when
linguistic var iables are used, these degrees may be managed by specif ic functions.
A fuzzy set is a pair ( A,m) where A is a set and .
For each , m( x) is the grade of membershi p of x. If A = { x1,..., xn} the fuzzy set ( A,m)
can be denoted {m( x1) / x1,...,m( xn) / xn}.
An element mapping to the value 0 means that the member is not included in the fuzzy set, 1
descr i bes a fully included member. Values str ictly between 0 and 1 character ize the fuzzy
members. The set is called the support of the fuzzy set ( A,m) and
the set is called the kernel of the fuzzy set ( A,m).
Fuzzy Logic as Human Logic
In reality exact rules that cover the respective case perfectly can only be def ined for a few
distinct cases. These rules are discrete points in the continuum of possi ble cases and humans
approximate them. This approximation, and likewise the abstraction and think ing in
analogies, are only rendered possi ble by the f lexi bility of µhuman logic.¶ To implement this
human logic in engineer ing solutions, a mathematical model is required. Fuzzy logic has been
developed as such a mathematical model. Fuzzy logic can be viewed as an extension of
multi-valued logic. The three-valued logic of Lukasiewicz, containing µtruth,¶ µintermediate¶
and false values, is considered to be the basis of fuzzy logic. Fuzzy reasoning is in fact a
reasoning that is neither exact nor absolutely inexact, but only to a cer tain degree exact or
inexact. Unlike the reasoning based on classical logic, fuzzy reasoning aims at the modeling
of reasoning schemes based on uncer tain or imprecise information. In fuzzy logic, individual
elements can be members of a set to only a cer tain degree, that is, they can belong in different
degrees to different sets simultaneously.
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 4/17
R etr ieved October 27, 2009, from Electronics for you website:
(htt p://www.electronicsforu.com/electronicsforu/Ar ticles/ad.asp?ur l=/efylinux/efyhome/cove
r/additions/fuzzy.htm&title=Fuzzy%20Logic%20and%20its%20Advantages)
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 5/17
Problems being addressed by methods
presented
Technology-dr iven research, where medicine provides AI with a good set of problems is being used to develop techniques leading to more general- purpose technologies. These can
then be applied to other domains (for example, diagnosis of computer circuits). Inward-
look ing research, which addresses technical solutions that, although neither directly AI- based
nor of immediate concern to clinicians, are needed to suppor t the long-term goals of both
enterpr ises.
One such problem is the design and implementation of integrated electronic patient records.
Inside Problems
The apparent lack of progress in AI in Medicine has been attr i buted to separate challenges
faced by AIM (Ar tif icial Intelligence in Medicine) researchers and medical professionals.
AIM researchers attempt to solve diff icult computational problems like acquisition and
representation of a practitioner¶s knowledge and sk ills, integration of episodic longitudinal
and histor ical patient data and communication with real time monitors and devices.
Medical professionals must acquire and process knowledge. They are also subject to human
factor. The impact of diagnosis and treatment on patient¶s life, the threat of professional
liability and pressures of administrative and f inancial constraints are other issues that def ine
the viewpoint of a medical practitioner. These differences have led to conf licting or
contradictory goals in the par tnershi p between AIM researcher and their medical
collaborators. AI has also been hampered by its own concerns and limitations. Handling the
large quantities of well established medical knowledge, adapting to the dynamic and
uncer tain nature of medical practice and discover ing and modelling how physicians
successfully perform diagnosis are some problems signif icantly related to the medical
community.
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 6/17
Ex ernal Problems
Some external problems have affected AI¶s performance and use in clinical settings. Two of
the most commonly cited are lack of user acceptance and lack of user infrastructure, the f irst
being a two way problem. Physicians are usually concerned about the computer tak ing over
the decision mak ing without substantiating the suggested decisions and that are frequently
based on very limited knowledge .the systems usually work on a small subset of the patient
cases. The systems usually work on interfaces that interfere with workf low of processing and
treating patients. This concern is signif icant when the ser iousness of the decisions is high.
The practitioners are still accountable for the decisions made and not the AI systems. People
are more stereotypical towards the old AI systems even though the new systems have
replaced the old ones and are much more accurate and useful.
Fuzzy Approaches by Becker et al. and
Yoshizawa et al.
Now let us see the different fuzzy approaches taken in the papers provided. (Becker,
H Kasmacher, Kalff, & Zimmerman, 1994) and (Yoshizawa, Takeda, Yambe, & Nitta, 1992).
Dur ing open-hear t surgery, management of narcosis and stabilisation of the patient¶s
circulation are the most impor tant tasks of the anaesthetist. Currently, he is suppor ted by
monitor ing devices which present vital parameters like blood pressures, temperatures and
blood gases. Conventional monitor ing devices are equi pped with threshold alarm facilities
which means that if any parameter exceeds a predetermined, f ixed threshold, an acoustical
and optical alarm signal is generated .The limitation of this approach is the diff iculty of
determining appropr iate thresholds for each signal and the fact that deter ioration of onehemodynamic parameter like loss of blood volume can lead to a change of all blood pressures
i.e. ar ter ial pressure, venous pressure, atr ial pressure etc., each tr igger ing a different alarm
signal.
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 7/17
Fig: Conventional Alarm system
Fig: Fuzzy Alarm system
This approach is an intelligent alarm system to suppor t the anaesthetist in monitor ing the
patients hemodynamic state dur ing open-hear t surgery. Based on the decision mak ing process
of the anaesthetist, the most impor tant vital parameter constellations are evaluated using a
fuzzy inference approach. The evaluation yields an estimation of f ive hemodynamic statevar iables for which a continuous alarm visualisation is presented on the user interface.
Dur ing the complicated hear t surgery the intelligent alarm system gathers all information
from Anaesthesia Information System (AIS). For processing of exper t knowledge in the form
of diagnostic rules a fuzzy-inference approach is well suited.
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 8/17
K nowledge acquisi ion: Membership Func ions
Ten anaesthetists who have been work ing in cardio-aesthesia for between 3 and 15 years
individually rated their meaning on the bandwidth of terms like: very low, a little too low,
good, a little too high and very high of the measured vital parameters. For every term each
anaesthetist checked a par t of the measurement scale.
Each of the n (=lo) anaesthetists received a weight w = l/n (l/l0) and the weights were
summed up for each term. These f indings were used to def ine a set of linear and symmetr ic
membershi p functions for the linguistic input var iables which descr i be the vital parameters
hear t rate, systolic ar ter ial pressure lef t atr ial pressure and central venous pressure.
Fig: Membershi p function for linguistic var iable: Ar ter ial systolic pressure
Fuzzy k nowledge base:
A fuzzy knowledge base for the evaluation of the hemodynamic state var iables sets of
questionnaires was created. For each hemodynamic state var iable 25 stimuli were presented
graphically using vir tual analogue scales for the parameter representation in order to avoid
bias caused by linguistic uncer tainty. These stimuli covered the whole hemodynamic state
space. Thir teen exper ienced cardio anaesthetists marked their evaluation of the state var iable
for each stimulus on the given analogue scale. This estimation was transformed into fuzzy
rules about the state var iable
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 9/17
Fig: Extraction of the fuzzy rule weights.
Fif ty rules were acquired for each state var iable in this way. These rules were implemented in
the knowledge base of the intelligent alarm system.
User interface
The intelligent alarms are presented on the lef t section of the user interface in a prof ilogram.
It displays information about the state var iables related to the evaluation of the vital
parameter constellations.
Vital Trend Visualisation ± VTV
In the r ight section of the user interface shows the trend of the most impor tant hemodynamic
vital parameters, The VTV trajectory enables the anaesthetist to recognise pathologic trends
in the patient's circulatory functions and enables him to prevent deter ioration of the patient's
state.
Explanation module
This visualisation allows the anaesthetist to comprehend the system evaluation and to predict
the behaviour of the state var iable af ter blood volume or drug application by visuallyextrapolating the trend curve.
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 10/17
Automatic monitor ing and Estimation
tool for the cardiovascular System:
Now Let us see the Automatic monitor ing and Estimation tool for the cardiovascular System.
(Yoshizawa, Takeda, Yambe, & Nitta, 1992) This was done under the ventr icular assistance
using fuzzy reasoning. To promote clinical use of the lef t ventr icular assist device (LVAD),
the medical specialist should be kept free from constant attention to the operation or the
control behaviour of the LVAD. In this study a system to monitor the LVAD control system
is developed. It is named TOTOMES. It has a function that detects the presence and or igin of
the accidents happening in the cardiovascular system under the LVAD assistance. The fuzzy
reasoning algor ithms are used to detect the malfunctions and control the LVAD operations.
In order to avoid the problems like infection, dr if ting and poor durability, invasive
measurements for detecting circulatory abnormalities and malfunctions of LVAD control
system was made as little as possi ble. Here the directly measured data are Instantaneous dr ive
pressure PDR V (t), ejecting dr ive pressure level PP , f illing dr ive pressure level P N, outf low
rate from the LVAD f AH(t), aor tic pressure p(t), ECG signal, reference stock volume of the
LVAD sv*and reference systolic duration of the LVAD
*. The data are stored every 10 ms
into the personal computer system where the TOTOMES was implemented. Then the direct
measurements are processed and stored in secondary var iables.
These data are stored every 10 ms into a PC. The parameters of cardiovascular dynamics are
identif ied in a real time fashion by a time ser ies model. The instantaneous value of cardiac
out put of natural hear t is also identif ied on the basis of identif ied parameters.
Detection Procedures
The objects of detection are as follows:
Hardware
o Air tube of Pneumatic dr iver
o Cannulae connected with the patient
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 11/17
o Aor tic Pressure sensor
Sof tware
o R unning away of sof tware for pneumatic controller
o Divergence of adaptive control algor ithm
Cardiovascular system
o Hear t rate
o Cardiovascular parameters
o Cardiac Out put
Detection of malfunctions are realised by application of fuzzy logic to secondary var iables.
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 12/17
R easons for choosing Fuzzy Logic
Fuzzy Logic reduces the design development cycle
Fuzzy Logic simplif ies design complexity
With a fuzzy logic design methodology some time consuming steps are eliminated.
Moreover, dur ing the debugging and tuning cycle you can change your system by simply
modifying rules, instead of redesigning the controller. In addition, since fuzzy is rule based,
you do not need to be an exper t in a high or low level language which hel ps you focus more
on your application instead of programming. As a result, Fuzzy Logic substantially reduces
the overall development cycle.
Fuzzy Logic improves time to mark et
As we explained above, a fuzzy based design methodology addresses the time on design
complexity and marketing the end product.
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 13/17
A Better Alternative Solution to Non-Linear Control
A linear approximation technique is relatively simple, however it tends to limit control
performance and may be costly to implement in cer tain applications. A piecewise linear
technique works better, although it is tedious to implement because it of ten requires the
design of several linear controllers. A lookup table technique may hel p improve control performance, but it is diff icult to debug and tune. Fur thermore in complex systems where
multi ple inputs exist, a lookup table may be impractical or very costly to implement due to its
large memory requirements.
Fuzzy logic provides an alternative solution to non-linear control because it is closer to the
real wor ld. Non-linear ity is handled by rules, membershi p functions, and the inference
process which results in improved performance, simpler implementation, and reduced design
costs
Fuzzy Logic improves control perf ormance
With fuzzy logic we can use rules and membershi p functions to approximate any continuous
function to any degree of precision. We can also add more rules to increase the accuracy of
the approximation (similar to a Four ier transform), which yields an improved control
performance. R ules are much simpler to implement and much easier to debug and tune than
piecewise linear or lookup table techniques.
Fuzzy Logic simplif ies implementation
A linear approximation requires handling each input separately which multi plies designeffor t. Similar ly, a piecewise linear approach requires the design of several controllers and is
costly to implement. A lookup table seems more appropr iate for this problem but it takes time
to develop, debug and tune. For example, if we assume that each input requires eight bits, a
lookup table would require 64K entr ies which makes it very time consuming to implement.
Fuzzy Logic reduces hardware costs
Conventional techniques in most real life applications require complex mathematical analysis
and modelling, f loating point algor ithms, and complex branching. This typically yields a
substantial size of object code which requires a high end DSP chi p to run. Fuzzy Logic
enables you to use a simple rule based approach which offers signif icant cost savings, both in
memory and processor class.
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 14/17
Adaptive Fuzzy Techniques
Conventional Fuzzy logic systems like the PID control systems cannot adapt to gradual
changes in environments but only can adjust the behaviour from one execution to another, but
the rules themselves cannot change. These can only be used where the environments are
known and predictable.
By including adaptive control in Fuzzy logic we can implement design systems that can
adjust to environmental changes. Physical systems are subject to long term permanent
changes due to wear and tear to physical par ts and sensors. To compensate this process
designers incorporate fault tolerance in their PID models or they can add an ancilliary system
which confers adaptive nature to the controllers.
Adaptive Fuzzy Logic
An adaptive Fuzzy Logic system adjusts to time or process phased conditions and changes
the suppor ting system controls. Thus an adaptive system modif ies the character istics of the
rules. It resembles neural network in the way they work. An adaptive fuzzy system is highly
sophisticated and has a higher degree of adaptive parameters.
An adaptive fuzzy controller (Chr istine M, Michael A, & Dimitr is K)
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 15/17
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 16/17
8/6/2019 Sysmodelling Shon Assignment 4
http://slidepdf.com/reader/full/sysmodelling-shon-assignment-4 17/17
BibliographyBecker, K., H Kasmacher, B., Kalff, R ., & Zimmerman, H.-J. (1994). A Fuzzy Logic Approach to
Intelligent Alarms in Cardioanesthesia. IEEE .
Chr istine M, H., Michael A, W., & Dimitr is K, P. Adaptive fuzzy controller that modif ies
membershi p functions.
Gaba, D., Maxwell, M., & De Anda, A. (1987). Anasthetic Mishaps: Break ing the chain of accident
Evolution. American Sociological Association.
Iserman, R . (1998). On fuzzy logic Applications for Automatic control, Supervision and Fault
diagnosis. IEEE .
Yoshizawa, M., Takeda, H., Yambe, T., & Nitta, S. -i. (1992). An Automatic Moniter ing and
Estimation tool for CardioVascular Assistance using Fuzzy reasoning. IEEE .