expert system for offline clinical guidelines and treatment · expert system for offline clinical...

Life Science Journal 2012;9(4) http://www.lifesciencesite.com

2639

Expert System for Offline Clinical Guidelines and Treatment

Tanzila Saba1, Saleh Al-Zahrani

2 and Amjad Rehman

1

1Faculty of Computing, Universiti Teknologi Malaysia (UTM), 81310 Skudai Johor Malaysia

2Faculty of Computer and Information Sciences, Al-Imam Muhammad bin Saud Islamic University Riyadh KSA

Abstract: Offline clinical guidelines are normally designed to coordinate a clinical knowledge base, patient

information and an inference engine to create case based guidance. In this regard, offline clinical guidelines are still

popular among the healthcare professionals for updating and support of clinical guidelines. Although their current

format and development process have several limitations, these could be improved with artificial intelligence

approaches such as expert systems/decision support systems. This paper first, presents up to date critical review of

existing clinical expert systems namely AAPHelpm, MYCIN, EMYCIN, PIP, GLIF and PROforma. Additionally,

an analysis is performed to evaluate all these fundamental clinical expert systems. Finally, this paper presents the

proposed clinical expert system design and development to help healthcare professionals for individual treatment.

[Saba T, Al-Zaharani S, Rehman A. Expert System for Offline Clinical Guidelines and Treatment Life Sci J

2012;9(4):2639-2658] (ISSN:1097-8135). http://www.lifesciencesite.com 393.

Keywords: Expert system, Quality of Information (QoI), Clinical guidelines, Diagnosis, Treatment.

1. Introduction

The developments of Expert Systems (ESs) have

drawn much attention of the research community in the

last few decades. The possibility of expert's systems

has additionally been generally acknowledged in

clinical environment, beginning from a straightforward

database inquiry to complex treatment proposal [1].

The logical writing gives the significant wellspring of

learning joined by nearby and rehearsed based proof.

The information of expert's systems exists as rules.

There are four zones during the time spent building up

a rule based experts system [2,4].

The main issue in the process of expert system

development is rule design and representation. This

territory concentrates on the representation of a rule,

such as type of expression, information sort, upkeep,

nearby adjustment, and so forth. The second territory is

rule securing, which is a procedure that encourages the

information obtaining process specifically from a space

experts. Thirdly, range is rule check & testing.

This procedure plans to guarantee the machine-

interpretable rule is unambiguous and grammatically

and also semantically right. Furthermore, we have to

test rules utilizing existing patient information. The last

procedure is rule execution, which concentrates on the

execution time and guarantees the rule motor can keep

running in different clinical spaces and in different

modes. Expert's systems appear to be extremely useful

in supporting clinical options; be that as it may, it is

hard to get wellbeing experts to really utilize them, so

Abbod et al. [3] abridges ten tenets for effective

execution of expert's systems in clinical environment.

These tenets are:

i. Speed of giving proposal

ii. Anticipate needs and convey progressively.

iii. Recognize that human services experts will

firmly oppose ceasing (recommendations) (need to

give elective choices to stay away from resistance)

iv. Changing bearing is simpler than ceasing.

v. Simple mediations work best a rule on a

solitary screen.

vi. Acquire extra data just when wanted.

vii. Monitor affect, get input and react.

viii. Oversee and keep up learning based systems.

Expert's systems are likewise utilized as a part of

different spaces of human services. For instance,

Chawla and Gunderman [5] compress reconnaissance

systems for early identification of bioterrorism-related

maladies. Coordination of Geographic Information

Systems (GIS) and human services permits depicting

and comprehension the changing spatial association of

social insurance, looking at the relationship between

wellbeing results and get to, and investigating how

conveyance of medicinal services could be enhanced

[6,7].The rest of the paper is organized further into

http://www.lifesciencesite.com/


2640

three main sections. Section 2, importance of AI in

clinical guidelines, section 3 presents in detail

background of computer interpretable guidelines,

section 4 describes clinical expert systems and their

merits/demerits; section 5 describes existing expert

systems for clinical guidelines along with their

shortcomings. Section 6 presents proposed expert

system. Finally, section 7 covers conclusion.

2. Artificial Intelligence and Clinical Guidelines

The history relation of A.I and clinical guidelines

is very old. The first practice started in early 80s.

Clinical guidelines are supposed to represent the best

clinical suggestions, records and practices.

Subsequently, are presently standouts amongst the

most focal ranges of research in Artificial Intelligence

(AI) in pharmaceutical and in medicinal basic

leadership. Clinical rules assume distinctive parts in the

clinical procedure: for instance, they can be utilized to

bolster doctors in the treatment of illnesses, or for

scrutinizing, for assessment, and for training purposes.

2.1. Collaborative decision making

Cooperative decision making is a circumstance

confronted when people all things considered settle on

a decision from the available options. This choice is no

more extended inferable from any single person who is

an individual from the gathering. This is on the

grounds that every one of the people and social

gathering procedures, for example, social impact adds

to the result. The choices made by gatherings are

frequently unique in relation to those made by people.

Bunch polarization is one clear illustration: bunches

tend to settle on choices that are more extraordinary

than those of its individual individuals, toward the

individual slants [11,12]. However, during the

application of clinical guidelines, there are moments

when this type of decision is required. The selection

among scientifically valid options during the medical

process must be done based on the opinions of the parts

involved (healthcare professionals and patients).

Technology-assisted decision making may help the

generation of ideas and actions, the choice of

alternatives and the negotiation of solutions. The

existence of CIGs and a tool for execution of medical

guidelines enables the implementation of automated

group decision making [13,14].

A group decision environment with a decision

model will help healthcare professionals and patients

clarify their position in the decision making process

and assure that their perspectives and preferences are

heard.

2.2. Expert Systems

A program that does a task that would some way

or another be performed by a human. For instance,

there are some expert systems that can analyze human

diseases, make money related gauges, and timetable

courses for conveyance vehicles. Some expert systems

are intended to replace human specialists, while others

are intended to help them [15,16].

Expert systems are a piece of a general class of

PC applications known as computerized reasoning. To

outline a specialist framework, one needs a learning

designer, a person who concentrates how human

specialists settle on choices and makes an interpretation

of the tenets into terms that a PC can get it [17].

Typical examples of clinical expert systems are

exhibited in Figure 1.


2641

Fig 1. Typical Expert Systems [14]


2642

2.3. Case-Based Reasoning (CBR)

Case-based reasoning (CBR), comprehensively

understood, is the way toward taking care of new

issues in light of the arrangements of comparative past

issues. An auto technician who settles a motor by

reviewing another auto that showed comparable side

effects is utilizing case-based thinking. A legal advisor

who advocates a specific result in a trial in view of

lawful points of reference or a judge who makes case

law is utilizing case-based thinking. In this, too, a

designer replicating working components of nature

(honing bio-mimicry), is regarding nature as a database

of answers for issues. Case-based thinking is an

unmistakable sort of similarity making [18,19].

It has been contended that case-based thinking is

an intense strategy for PC thinking, as well as an

unavoidable conduct in regular human critical thinking;

or, all the more drastically, that all thinking depends on

past cases by and by experienced. This view is

identified with model hypothesis, which is most

profoundly investigated in psychological science [20-

22].

Case-based thinking has been formalized for

motivations behind PC thinking as a four-stage

process[23]:

Recover: Given an objective issue, recover from

memory cases pertinent to settling it. A case comprises

of an issue, its answer, and, normally, comments about

how the arrangement was inferred. For instance,

assume Fred needs to get ready blueberry flapjacks.

Being a beginner cook, the most important experience

he can review is one in which he effectively made plain

hotcakes. The strategy he took after for making the

plain hotcakes, together with supports for choices made

en route [24].

Reuse: Map arrangement from past case to

objective issues. This might include adjusting

arrangement as expected to fit the new circumstance. In

the flapjack illustration, Fred must adjust his recovered

answer for incorporate the expansion of blueberries.

Reconsider: Having mapped the past answer for

the objective circumstance, test the new arrangement in

this present reality (or a reproduction) and, if essential,

change. Assume Fred adjusted his flapjack

arrangement by adding blueberries to the player. In the

wake of blending, he finds that the player has turned

blue, an undesired impact. This recommends the

accompanying amendment: defer the expansion of

blueberries until after the player has been scooped into

the container [25].

Hold: After arrangement has been effectively

adjusted to objective issue, store subsequent experience

as another case in memory. Fred, as needs be, records

his recently discovered strategy for making blueberry

flapjacks, in this manner improving his arrangement of

put away encounters, and better setting him up for

future hotcake making requests. The average Case

Base Reasoning is shown in Figure 2.

3.Computer-Interpretable Guidelines

Clinical rules are suggestions on the fitting

treatment and care of individuals with particular

maladies and conditions. Clinical rules depend on the

best accessible confirmation. There is an arrangement

of components that rule scientists might want to see

rules get [26]. The goal of scientists is to change the

static and uninvolved nature of rules. Clinical rules

assume diverse parts in the clinical procedure: for

instance, they can be utilized to bolster doctors in the

treatment of sicknesses, or for scrutinizing, for

assessment, and for training purposes. A wide range of

systems and activities have been produced as of late

with a specific end goal to acknowledge PC helped

administration of clinical rules [27,28].


2643

Fig 2. Typical Case Based Reasoning system [31]

4. Clinical Expert Systems

4.1. What are Clinical Expert Systems?

Clinical expert systems help human experts in

their work; be that as it may, they don't supplant their

insight and abilities. Basic components of these

systems that are intended to give quiet particular

direction incorporate the learning base (e.g., ordered

clinical data on analyses, sedate associations, and

rules), a program for joining that learning with patient-

particular data, and a correspondence instrument—at

the end of the day, a method for entering tolerant

information into the application and giving significant

data (e.g., arrangements of conceivable judgments,

tranquilize connection alarms, or preventive care

updates) back to the clinician [29,30].

Expert systems could be actualized utilizing an

assortment of stages (e.g. web based, neighbourhood

PC, organized EMR, or a handheld gadget).

Additionally, an assortment of figuring methodologies

can be utilized. These methodologies may rely on upon

whether the expert's systems is incorporated with the

neighbourhood EMR, whether current information is

accessible from focal archive or whether whole

framework is housed outside neighbourhood site and is

gotten to, yet not fused into neighbourhood EMR. On a

fundamental level, any sort of expert's systems could

use any of these basic computational models,

techniques for get to, or gadgets. The decisions among

these components may depend more on the kind of

clinical systems as of now set up, merchant offerings,

work process, security, and monetary limitations than

on the sort or reason for the experts systems [31-33].

4.2. Significance of Clinical Expert Systems


2644

Clinical expert's systems are methodically created

articulations to help medicinal services experts and

patients about clinical conditions [34]. At first, rules

are just in light of the agreement of gatherings of

specialists; however with the development of

confirmation based clinical practice, different methods

are incorporated into rule improvement. The Delphi

and Ostensible gathering strategies are a portion of the

procedures that are later incorporated into the

improvement procedure are still utilized today.

Presently, clinical expert systems advancement is

focused on around a broad research of the writing and

exhaustive examination of experimental proof. The

procedure typically begins with the decision of the rule

theme or subject, in light of the issues that persuade the

advancement. Clinical rules can be created to an

extensive variety of subjects and clinical territories,

including wellbeing conditions bound to illnesses and

efficient expenses. To pick the subject, it is important

to do a preparatory check of the accessible

confirmation keeping in mind the end goal to find out

the legitimacy of the topic [35-38].

A large number of the innovation contrasts

depicted in the past few decades. The accompanying

elements might be more pertinent to clinician client /

helping in execution:

(1) Essential needs or issue and objective territory

of look after which experts is being considered (e.g.,

enhance general productivity, distinguish ailment early,

guide in precise analysis or convention based

treatment, or avoid hazardous antagonistic occasions

influencing the patient);

(2) Main choice is whether experts could consider

need or issue distinguished patient from others [40-43].

Experts can give support to clinicians at different

stages in the care procedure, from preventive care

through determination and treatment to checking and

development. Experts as actualized today could

incorporate, for instance, arrange sets custom-made for

specific conditions or sorts of patients (preferably in

light of confirmation based rules and tweaked to reflect

individual clinicians' inclinations), access to rules and

other outside databases that can give data applicable to

specific patients, updates for preventive care, and

alarms about conceivably perilous circumstances that

should be addressed[44,45].

The most well-known utilization of experts is for

tending to clinical needs, for example, guaranteeing

exact findings, and screening in a convenient way for

preventable sicknesses, or turning away unfriendly

medication occasions. In any case, experts can likewise

conceivably bring down expenses, enhance

productivity, and decrease understanding burden. Truth

be told, experts can now and then address each of the

three of these ranges at the same time. For instance, by

alarming clinicians to possibly duplicative testing. For

more mind boggling intellectual assignments, for

example, demonstrative basic leadership, the point of

experts is to help, as opposed to supplant, the clinician

though for different errands, the experts may calm

clinician of the weight of remaking requests for every

experience. The experts may offer proposals, yet

clinician must channel data, survey recommendations,

& choose whether make a move / what move to make?

[46,47].

4.3 Expert's Impact

Current area concentrates on assessments of the

effect of experts on human services quality, utilizing

Donabedian's great meaning of value including

structure, process, and results of medicinal services.

Donabedian upheld that authoritative results, for

example, cost and productivity, and in addition

singular patient wellbeing results, be assessed.

Donabedian's model is extended via Carayon and her

associates' plan of structure, which incorporates

individuals, association, innovations, assignments, and

environment. This extended meaning of structure is

utilized here with the goal that experts affect on cost

and productivity are tended to and included as a major

aspect of effect on structure [48,49]. As delineated

underneath, assessment of effect incorporates mind

process and patient wellbeing results. Basic results are

likewise tended to beneath.

Most distributed assessments of the effect of

experts on medicinal services quality have been

directed in inpatient as opposed to mobile settings, and

most have been in huge scholarly therapeutic focuses,

frequently utilizing "homegrown" systems, where there

is a culture that is acclimated to their utilization and

satisfactory assets (counting ability, time, and


2645

foundation) to fabricate and look after them. Albeit

numerous business EMRs have expert's capacities,

there has been minimal orderly research on the results

or even on the execution procedures of business

experts in group settings. These oversights, and the

restricted concentration in charge of them, are

especially risky since most doctors' facilities will send

business systems later on, and their way of life and

assets are probably going to vary from those of

expansive scholastic therapeutic focuses. Moreover, the

effect of experts in mobile settings needs more

consideration. A portion of the undertakings inside the

AHRQ Ambulatory Safety and Quality Program are

starting to address this need [50-52].

The exploration on experts has other critical

confinements. To start with, in spite of the fact that

various experts' suggestions have been distributed,

similarly, most research has inspected impacts of

experts on the procedure of care (as opposed to the

results or structure) and has concentrated principally on

clinician basic leadership. Finally, indicative projects

have had restricted use by and by settings. At last, after

effects of examination to date are blended as far as

viability of experts for specific conditions or specific

sorts of experts. These impediments indicate holes in

the writing. In spite of the fact that RCTs are viewed as

the best quality level for research concentrates on,

subjective studies might be better ready to decide why

experts mediation succeeds or comes up short [53].

The accompanying area audits the after effects of

RCT studies and different investigations of experts.

Since the greater part of studies manage process and

patient wellbeing results, these viewpoints is talked

about initially, trailed by an exchange of structure [54].

4.4 Impact on Care Process and Patient Health

Outcomes

The meta-investigations of alarms and updates for

choice support have been genuinely reliable in

demonstrating that they can change clinician basic

leadership and activities, diminish drug blunders, and

advance preventive screening and utilization of proof

based proposals for pharmaceutical solutions. The

information on how those choices influence tolerant

results are more restricted, in spite of the fact that

various studies have demonstrated beneficial outcomes.

By and large, the outcomes demonstrate the capability

of experts to enhance the nature of care [55,56].

In spite of the fact that the studies demonstrating

the capacity of experts to anticipate prescription

mistakes (off base choices) have been reliably positive,

the consequences of research studies on the capacity of

experts to turn away unfavourable medication

occasions (mischief to the patient) have had a tendency

to be blended. Few of the studies looking at the effect

on wellbeing results are RCTs, many studies are

ineffectively outlined, and not all studies indicated

factually critical impacts. Expert's concentrates on that

concentrate on giving analytic choice support have

likewise demonstrated blended results, and less of

these systems have been assessed practically speaking

settings. In any case, contemplates contrasting experts

demonstrative proposals and experts clinicians'

examinations of testing clinical cases have

demonstrated that the symptomatic experts can help

even experts doctors to remember conceivably

imperative analyses they didn't at first consider [57,58].

A portion of the blended results have come about

because of methodological issues, for example, roof

impacts (execution is as of now great preceding

actualizing experts) or low factual energy to recognize

measurably.

Fig 3: The PROforma task model

Plans are the essential building pieces of a rule &

might composed of any number of undertakings of any

sort, including different arrangements. Options are

taken at focuses where alternatives are introduced, such

as whether to treat a patient or complete further

examinations. Activities are regularly clinical

methodology, (for example, the organization of an

infusion) which should be completed. Enquiries are

commonly asks for additional data or information,


2646

required before the rule can continue. PROforma

programming comprises of a graphical editorial

manager to bolster the creating procedure, and a motor

to execute the rule particular. The motor can likewise

be utilized as an analyzer amid the application

improvement stage. A PROforma editorial manager

bolsters the development of a rule as far as the four

undertaking sorts. Utilizing the symbols appeared as a

part of the outline, systems of errands can be created

that speak to arrangements or strategies did after some

time. In the manager, legitimate and worldly

connections between assignments are caught normally

by connecting them as required with bolts. Any

procedural and restorative learning required by the rule

all in all or by an individual assignment is entered

utilizing formats connected to every errand. The

subsequent populated graphical structure is naturally

changed over into a database prepared for execution

[85-86].

4. 5. GLARE

GLARE stands for Guide Line Acquisition,

Representation and Execution. Following an analysis

of some of the existing expert systems, it is observed

that this area of research is still fresh. Due to these

various shortcomings, none of the models is largely

adopted by the health informatics community. The

level of complexity is another issue to be addressed

[87]. Table 1 presents an up-to-date review of the

existing expert systems.

The maintenance and library partitions possess

administrative information about the guideline, namely

authoring and version number. The constructs of the

maintenance partition are title, (file) name, author,

version, institution, date of last modification and

validation status. The validation status contains

information about the approval of the guideline in a

local institution and it may have three possible values:

testing, research, production and expired. The

transition from testing to production means a shift of

responsibility from the institution that developed the

MLM to the local institution where the guideline will

be applied. The library compartment contains

constructs used for a detailed description of the

guideline and among them the attribute purpose

enables the expression of the medical objective of the

MLM [88].

The main constructs of the knowledge

compartment are data, evoke, logic and action. The

data construct is used to obtain the values of the

concepts referred in the MLM from the information

system of the healthcare institution. The evoke

construct contains the events that execute MLM and

such events are related with the medical parameters in

data. The decision criteria are expressed in the logic

construct through if-then-else rules and sets of logical,

mathematical and temporal operators. When a rule is

assessed to the value true, a given procedure of the

construct action is proposed. These procedures may

include messages/alerts or the execution of other

MLMs. This approach reveals great modularity and

gives transparency to the decision making process, but

given its simplicity, the ability to capture the full

content of a medical guideline is compromised. Arden

Syntax is mainly used in alert and monitoring systems,

like the ones provided by the Regenstrief Institute. It is

defined in Backus-Naur Form (BNF), a notation

technique used to describe the syntax of computation

languages. Currently the development of Arden Syntax

by HL7 is based on XML [89,90].


2650

Table 1 Existing Expert Systems for clinical guideline development

5. Proposed Clinical Expert System

An expert system is a program that prevails upon

information of some experts with a view to taking care

of issues or giving exhortation. The way toward

building expert system is frequently called information

designing. The learning designer is included with all

parts of a specialist framework. Building expert system

is for the most part an iterative procedure. The parts

and their association will be refined through the span of

various gatherings of the information build with the

experts and clients system. A general structure of the

proposed expert system is exhibited in Figure 4.

Fig 4. Proposed Expert System’s Structure

5.1 Knowledge

Hence, we established the following knowledge

acquisition procedure with the assistance of our expert

(a physician).

Examine all guidelines to be implemented and create

variable nodes. Most of the time an inclusion,

exclusion criterion would generate a node itself.

Complex criteria could be broken down to simpler

random variables. Superficial variables can also be

created that will act as intermediate nodes to combine

related variables. Find all the common nodes to

Expert Systems Features Decision

Model

Standards adopted

AAPHelp

MYCIN

Rule based

Decision support system

(DSS)

Bayesian

approach

Rule based

No

XML

EMYCIN

DSS

Rule-based

PUFF

GLIF

Decision based on

patient condition

Event-based

and rule-based

RDF,HL7

PIP

Query, plan, decision,

Rule-based

UMLS

PROforma Context based decision Event based XML,

GLARE

Query based decision

Rule-based

XML,

ICD-9


2652

protocols and draw arcs between them where there is a

cause and effect relationship [92].

We asked the expert to provide as many conditional

and prior probabilities as possible. Look for nodes in

individual inclusion�exclusion criteria that can cause

the common nodes or are effects of the common nodes

and draw appropriate arcs between them. Group the

criteria using the model shown in Figure 5.

Fig 5. Knowledge acquisition process [91]

5.2. Proposed Decision Model

Expert systems have also been neglected in the

current medical approaches. The current systems do

not complement the decision schemes proposed by

their models with techniques to infer the confidence in

the outputs of the decision process. Furthermore, the

problem of incomplete and uncertain information

mentioned in previous sections of this paper remains

unaddressed [93,94].

Expert systems use knowledge and inference

procedures to solve difficult problems. They have to

mimic the adaptation capabilities of human beings in

order to find solutions to new problems [95]. There are

four fundamental components of expert system [96]:

knowledge acquirement part, knowledge base,

inference engine and an interface. The knowledge

representation in expert system applies concepts of

logics to create structured formalisms and inference

rules. The knowledge itself may be introduced in the

system by human experts as rules, obtained from past

experience through learning algorithms or both [97].

The proposed system structure is easy to

understand. An interface manager passes user requests

to the inferencing mechanism and allows access to the

knowledge base. It also stores information about the

state of the session. The inferencing mechanism


2652

requests the hypergraph manager, which builds the

hypergraphs it needs. The classes that make up these

subsystems are all constructed from a number of basic

classes, such as linked lists, hash tables and matrices.

The proposed expert system is not composed

purely and simply of classes, but of collections of

classes that form sub-systems. The separate subsystems

could be altered internally without affecting the others,

as long as their interface to those others is unchanged.

The interfaces evolve quickly as the user demands new

features. However, if the interfaces are kept small the

efforts of change management is localized [98].

The system is not hierarchical in the sense that a

structured program is. Although the interface manager

could be described as a top level function, it performs

many functions and these may be changed. Hence, the

proposed expert system is exhibited in Figure 3.

This loose and simple organization would not

have resulted from a top-down procedure-oriented

methodology. It is therefore, possible to remove any

subsystem and replace it. For example, the Hypergraph

Manager and Inferencing Mechanism could be

removed and replaced with an ordinary backtracking

mechanism [99].

A new Interface Manager could be designed and

fitted onto the rest of the system, without changing

other subsystems. It has made the production of

successive versions easy. Strength of the system is that

it produced classes which could be reused in other

applications, almost as a side effect. Knowledge base

and interface classes might be reusable in other

applications. These properties mean that successive

versions of the system could be developed without

having to start from scratch. This is important because

new demands from users continually arise from their

experience with the system. Additionally, object-

oriented programming languages prefer to procedural

languages because they have inheritance mechanisms.

Inheritance allows one class to include another as part

of its definition. Unwanted plans could be over fitted

and new parameters could be added. The inheriting

class is known as the subclass and the parent class the

super class. A super class may be virtual i.e. it cannot

be used to create objects, as it does not fully defined

one. It serves only as a repository for definitions and

codes that subclasses share.

This decision model leaves the door open for

further research on the complementarily that other

techniques that manage uncertainty, namely Bayesian

Networks, Dempster-Shafer Theory and fuzzy Logic,

may offer to the QoI, since they manage uncertainty in

different ways. The implementation of such a decision

model is essential in order to capture context of the

execution of guidelines and provide measures of

confidence in the outputs [100].

The development of such a decision model is but

a step in the construction of a wider decision platform,

represented in Fig. 6, where healthcare professionals,

members of the same medical team, possibly dispersed

across different locations, can discuss de case of a

patient in the context of an intelligent environment.

Through the use of AI techniques it is possible to

perceive information about the state of stakeholders,

namely their attitudes and emotions and thus determine

the type of interaction they are developing. If one

throws into the equation relevant knowledge, from

exterior sources, concerning the health condition (that

is the object of the discussion) and guideline

recommendations, a group decision environment is

established for healthcare professionals to discuss if a

guideline is suitable for the situation at hand and

mediate/negotiate solutions. Having all this

information enables the medical team, in cases of non-

compliance of guidelines, to build new strategies and

adapt their content to maximize the probability of a

successful treatment.


2653

Fig 6. Decision model for the execution of clinical guidelines in proposed Expert System.

The development of such a decision model is but a

step in the construction of a wider decision platform,

represented in Fig. 3, where healthcare professionals,

members of the same clinical team, possibly

dispersed across different locations, can discuss de

case of a patient in the context of an intelligent

environment [101].

Through the use of AI techniques it is possible

to perceive information about the state of

stakeholders, namely their attitudes and emotions and

thus determine the type of interaction they are

develop knowledge, from exterior sources,

concerning the health condition (that is the object of

the discussion) and guideline recommendations, a

group decision environment is established for

healthcare professionals to discuss if a guideline is

suitable for the situation at hand and

mediate/negotiate solutions. Having all this

information enables the clinical team, in cases of

non-compliance of guidelines, to build new strategies

and adapt their content to maximize the probability of

a successful treatment [102-103].

6. Conclusion

There are several benefits of using expert

systems to assist healthcare professionals in order to

cure patients. Initially, systems knowledge is

autonomic. Thus, healthcare professionals do not

need to put enough efforts, labour in constructing and

knowledge base updating. When constructing or

maintaining the knowledge source, the only desired

action is to train the model with data. Secondly, the

extracted knowledge is the real world better. Hence,

this paper has presented decision support system to

help healthcare professionals for efficient clinical

treatment. This is an ongoing research line with

several experts working on the implementation of

useful expert systems and the development of

execution engines.

It is also observed that there is a need for a

decision model that addresses the aspects of the

contextualization of guideline execution and the

handling of incomplete and uncertain information.

However, opinions of healthcare professionals have

their own worth and should be included in the whole

process of development of knowledge base of the

expert system. Additionally, one of the current issues

of the guidelines is that they are too rigid and do not

give space for innovation and change. This decision

model of the expert systems should be flexible to

incorporate up-to-date changes.


2654

References

1. Gardner R, Pryor T, Warner H (1999). The HELP

hospital information system: update 1998. Int J of Med

Inform, 54(3), 169-182.

2. Gruber T (1993). A translation approach to portable

ontologies. Knowl Acquis 5(2):199-220.

3. Abbod M, von Keyserlingk D, Linkens D, Mahfoul M

(2001) Survey of utilization of fuzzy technology in

Medicine and Healthcare. Fuzzy Sets Syst 120:331-349

4. Haron, H. Rehman, A. Wulandhari, L.A. and Saba, T.

(2011) Improved Vertex Chain Code Algorithm for

Curve Length Estimation‖ Journal of Computer Science,

vol. 7(5): 736-743. doi: 10.3844/jcssp.2011.736.743

5. Chawla A, Gunderman R (2008) Defensive medicine:

prevalence, implications, and recommendations. Acad

Rad 15 (7): 948-9

6. Househ M, Kushniruk A, Maclure M, Carleton B,

Cloutier-Fisher D (2011) The use of conferencing

technologies to support drug policy group knowledge

exchange processes: An action case approach. Inter J

Med Inform 80: 251-261

7. Hripcsak G (1994) Writing Arden Syntax clinical logic

modules. Comput Biol Med 24(5): 331-363.

8. Kurniawan, F. Rahim,MSM, Daman, D. Rehman, A.

Mohamad, D. and Mariyam, S. (2011). Region-based

Touched Character Segmentation in Handwritten Words.

International Journal of Innovative Computing,

Information and Control vol. 7(6), 3107-3120.

9. Kurniawan, F. Rehman, A. Mohamad, D. (2009a). From

Contours to Characters Segmentation of Cursive

Handwritten Words with Neural Assistance. IEEE

International Conference on Instrumentation,

Communications, Information Technology and

Bioclinical Engineering (ICICI-BME), 12-18.

10. Karacapilidis N, Pappis C (1997) A framework for

group Expert systems: Combining AI tools and OR

techniques. Eur J Oper Res 103: 373-388.

11. Bichindarits I, Marling C (2006) Case-based reasoning

in health sciences: What’s next? Artif Intell Med 36:

127-135

12. Bodenreider O (2006) The Unified Clinical Language

System (UMLS): integrating bioclinical terminology.

Nucleic Acids Res 32: 267-270

13. Bottrighi A, Terenziani P, Montani S, Torchio M,

Molino G (2006) Clinical guidelines contextualization in

GLARE. AMIA Annu Sympos Proc: p 860

14. Novais P, Carneiro D, Gomes M, Neves J (2012). Non-

invasive Estimation of Stress in Conflict Resolution

Environments. In Demazeau Y, Müller J, Corchado J,

Bajo J (eds.) Advances on Practical Applications of

Agents and Multi-Agent Systems - 10th International

Conference on Practical Applications of Agents and

Multi-Agent Systems (PAAMS 2012) Springer - Series

Advances in Intelligent and Soft Computing, vol 155, pp

153-160

15. Boxwala A, Peleg M, Tu S, Ogunyemi O, Zeng Q,

Wang D, Patel V, Greenes R, Shortliffe E (2004) GLIF3:

a representation format for sharable computer-

interpretable clinical practice guidelines. J Biomed

Inform 37: 147-161

16. Fox J, Black R, Chronakis I, Dunlop R, Patkar V, South

M, Thomson R (2008) From Guidelines to Careflows:

Modelling and Supporting Complex Clinical Processes.

In: Teije A, Miksch S, Lucas P (eds) Computer-based

Clinical Guidelines and Protocols: A Primer and Current

Trends. IOS Press, Amsterdam, pp 44-62

17. Melle W (1978) MYCIN: a knowledge-based

consultation program for infectious disease diagnosis.

Int J Man-Mach Stud 10(3): 313-322.

18. Cheater F, Closs S (1997). The effectiveness of methods

of dissemination and implementation of clinical

guidelines: a selective review. Clin Eff Nurs 1: 4-15

19. Seto E, Leonard K, Cafazzo J, Barnsley J, Masino C,

Ross H (2012) Developing healthcare rule based expert

systems: Case study of a heart failure telemonitoring

system. Inter J Med Inform.

doi:10.1016/j.ijmedinf.2012.03.001.

20. Saba, T. and Rehman, A.(2012). Effects of Artificially

Intelligent Tools on Pattern Recognition, International

Journal of Machine Learning and Cybernetics. DOI:

10.1007/s13042-012-0082-z.

21. Bradburn C, Zeleznikow J (1994) The application of

case-based reasoning to the tasks of health care

planning. In: Wess S, Althoff K, Richter M (eds) Topics

in case-based reasoning: first European Workshop,

EWCBR-93. Springer-Verlag, Berlin pp 365-78


2655

22. Bekhti, S. Rehman, A. Al-Harbi, M., Saba, T. (2011).

AQuASys An Arabic Question-Answering System based

on Extensive Question Analysis and Answer Relevance

Scoring. Information and Computation International

Journal of Academic Research, 3(4): 45-54.

23. Schalkoff R (1990) Artificial Intelligence: an

engineering approach. McGraw-Hill, New York.

24. Bratko I, Mozetiec I, Lavraec N (1989) KARDIO: A

study in deep and qualitative knowledge for ESs. MIT

Press, Cambridge

25. Elkin P, Peleg M, Lacson R, Bernstam E, Tu S, Boxwala

A (2000) Toward standardization of electronic guideline

representation. MD Comput 17(6): 39-44

26. Fox J, Alabassi A, Patkar V, Rose T, Black E (2006) An

ontological approach to modelling tasks and goals.

Comput Biol Med 36: 837–856

27. Isern D, Sánchez D, Moreno A (2011) Ontology-driven

execution of clinical guidelines. Comput Methods Prog

Biomed. doi: 10.1016/j.cmpb.2011.06.006

28. Jackson P (1990) Introduction to ESs. Addison-Wesley

Longman Publishing, Boston

29. Kaiser K, Miksch S (2009) Versioning computer-

interpretable guidelines: Semi-automatic modelling of

„Living Guidelines‟ using an information extraction

method. Artif Intell Med 46(1): 55-66.

30. Kalra J (2004) Clinical errors: an introduction to

concepts. Clinical biochemistry 37 (12): 1043-51.

31. Kavanagh B (2009) The GRADE System for Rating

Clinical Guidelines. PLoS Med 6 (9): 5.

32. Konglingren L (2009) HL7: Health Level Seven.

Interface:1-23

33. Koton P (1988) Reasoning about evidence in causal

explanations. In: Mitchell T, Smith R (eds). Proc AAAI-

88. AAAI Press, Menlo Park, 256-61.

34. Clark D (1990) Numerical and symbolic approaches to

uncertainty in AI. Artif Intell Rev 4: 109- 146.

35. Codish, S, Shiffman R (2005) A Model of Ambiguity

and Vagueness in Clinical Practice Guideline

Recommendations. AMIA Annu Symp Proc: pp 146-150

36. Johnson C, Turley J (2006) The significance of cognitive

modeling in building healthcare interfaces. Int J Med

Inform 75: 163-172.

37. de Clercq P, Blom J, Korsten H, Hasman A (2004)

Approaches for creating computer-interpretable

guidelines that facilitate decision support. Artif Intell

Med 31(1): 1-27

38. Haron, H. Rehman, A. Adi, DIS, Lim, S.P. and Saba,

T.(2012). Parameterization Method on B-Spline Curve.

Mathematical Problems in Engineering, 2012,

doi:10.1155/2012/640472.

39. Dennis S, Edwards S, Partridge M, Pinnock H, Qureshi

S (2004) The dissemination of the British guideline on

the management of Asthma. Respir Med 98: 832-837.

40. Elarbi-Boudihir, M. Rehman, A.(2011), T.Saba. Video

Motion Perception Using Operation Gabor Filter‖.

International Journal of Physical Sciences,6(12): 2799-

2806.

41. Saba, T. Rehman, A. and Elarbi-Boudihir, M. (2011b).

Methods and Strategies on off-line Cursive Touched

Characters Segmentation: A Directional Review.

Artificial Intelligence Review DOI 10.1007/s10462-011-

9271-5.

42. Samwald M, Fehre K, Bruin J, Adlassnig K-P (2012)

The Arden Syntax standard for clinical decision support:

Experiences and directions. J Biomed Inform.

doi:10.1016/j.jbi.2012.02.001

43. Schadow G, Mead C, Walker D (2006) The HL7

reference information model under scrutiny. Stud Health

Technol Inform 124: 151-156.

44. Schalkoff R (1990) Artificial Intelligence: an

engineering approach. McGraw-Hill, New York.

45. Shenoy P, Shafer G (1986) Propagating belief functions

with local computations. IEEE Expert 1(3): 43-52.

46. Shiffman R, Karras B, Agrawal A, Chen R, Marenco L,

Nath S (2000) GEM: a proposal for a more

comprehensive guideline document model using XML. J

Am Med Inform Assoc (5):488-498.

47. Aamodt A, Plaza E (1994) Case-Based Reasoning:

Foundational Issues, Methodological Variations, and

SystemApproaches. AI Commun 7(1): 39-59

48. Antoniou G, Harmelen F (2009) Web Ontology

Language: OWL. In: Staab S, Studer R (eds) Handbook

on Ontologies. Springer Heidelberg, Berlin, pp 91-110

49. Novais P, Salazar M, Ribeiro J, Analide C, Neves J

(2010) Decision Making and Quality-of- Information.

In: Corchado E, Novais P, Analide C, Sedano J (eds)

Soft Computing Models in Industrial and Environmental

Applications, 5th International Workshop (SOCO 2010).


2656

50. Springer - Series Advances in Intelligent and Soft

Computing vol. 73, pp 187-195

51. Oliveira T, Costa Â, Novais P, Neves J (2012) An

Interpretable Guideline Model to Handle Incomplete

Information. In: Omatu S, Paz Santana J, González S,

Jose M, Bernardos A, Corchado J (eds) Distributed

Computing and Artificial Intelligence, 9th International

52. Conference (DCAI 2012), Springer - Series Advances in

Intelligent and Soft Computing, pp: 437-444.

53. Ohno-Machado L, Gennari J, Murphy S, Jain N, Tu S,

Oliver D et al. (1998) The guideline interchange format.

J Am Med Inform Assoc: 5(4): 357-372.

54. Pearl J (1986) Fusion, propagation, and structuring in

belief networks. Artif Intell 29(3): 241-288.

55. Peleg M, Tu S, Bury J, Ciccarese P, Fox J, Greenes RA,

Hall R, Johnson PD, Jones N, Kumar A, Miksch S,

Quaglini S, Seyfang A, Shortliffe EH, Stefanelli M

(2003) Comparing computer interpretable guideline

models: a case-study approach. J Am Med Inform Assoc

10(1):52-68.

56. Bareiss R (1989) Exemplar-based knowledge

acquisition: a unified approach to concept

representation, classification, and learning. Academic

Press, San Diego.

57. Berger J (1994) Roentgen: radiation therapy and case-

based reasoning. In: O‟Leary D, Selfridge P (eds) Proc

10th Conf Artif Intell Appl. IEEE Computer Society

Press, Los Alamitos, pp 171-177.

58. Berners-Lee T, Hendler J, Lassila O (2001) The

Semantic Web. Sci Am 284(5): 34-43

59. Kurniawan, F. Rehman, A. Mohamed, D (2009b).

Contour Vs Non-Contour based Word Segmentation

from Handwritten Text Lines. An Experimental

Analysis. International Journal of Digital Content

Technology and its Applications, vol. 3(2), 127-131.

60. Logan R, Scott P (1996) Uncertainty in clinical practice:

implications for quality and costs of health care. The

Lancet 347: 595-98.

61. Mead, P. (2000) Clinical guidelines: promoting clinical

effectiveness or a professional minefield? J Adv Nurs 31

(1): 110-116.

62. Miller M, Kearney N (2004) Guidelines for clinical

practice: development, dissemination and

implementation. Int J Nurs Stud 41 (7): 813-821

63. Mohamad, I. Rahim, MSM, Bade, A. Rehman, A. and

Saba, T.(2012). Enhancement of the Refinement Process

for Surface of 3D Object. Journal of American Science.

8(4):358-365.

64. Neves J, Ribeiro J, Pereira P, Alves V, Machado J,

Abelha A, Novais P, Analide C, Santos M, Fernndez-

Delgado M (2012) Evolutionary intelligence in asphalt

pavement modeling and quality-of-information. Prog

Artif Intell 1(1): 119-135.

65. Lucas P (2004) Bayesian networks in biomedicine and

health-care. Artif Intell Med 30(3): 201-214.

66. Nikolopoulos (1997) Expert systems: Introduction to

First and Second Generation and Hybrid Knowledge

Based Systems. CRC Press, New York.

67. Dempster A (1967) Upper and lower probability

inferences based on a sample from a finite univariate

population. Biometrika 54(3-4): 515.

68. Peleg M, Wang D (2006) The Guideline Interchange

Format. Open Clinical.

http://www.openclinical.org/gmm_glif.html. Accessed

23 May 2012.

69. Perner P (1999) An architecture for a CBR image

segmentation system. Intern J Med Inform 12(6): 749-

59.

70. Quaglini S, Stefanelli M, Cavallini A, Micieli G, Facino

C, Mossa C (2000) Guideline-based careflow systems.

Artif Intell Med. 20(1): 5-22.

71. Rahim, MSM. Rehman, A. Faizal-Ab-Jabal, M., Saba, T.

(2011). Close Spanning Tree Approach for Error

Detection and Correction for 2D CAD Drawing.

International Journal of Academic Research, 3(4):

pp.525-535.

72. Rahim, MSM, Rehman, A. Sholihah, N. Kurniawan, F.

and Mohamad, D. (2012) Region-based Features

Extraction in Ear Biometrics, International Journal of

Academic Research,4(1): 37-42.

73. Ram P, Berg D, Tu S, Mansfield G, Ye Q, Abarbanel R,

et al. (2004) Executing clinical practice guidelines using

the SAGE execution engine. Medinfo 11(1):251-255.

74. Rehman, A. (2012). Machine Learning based Air Traffic

Control Strategy, International Journal of Machine

Learning and Cybernatics DOI: 10.1007/s13042-012-

0096-6.


2657

75. Rehman, A. and Mohamad, D. (2008). A Simple

Segmentation Approach for Unconstrained Cursive

Handwritten Words in Conjunction of Neural Network.

International Journal of Image Processing, 2(3): 29-35.

76. Rehman, A. and Saba, T. (2011a). Performance Analysis

of Segmentation Approach for Cursive Handwritten

Word Recognition on Benchmark Database. Digital

Signal Processing, 21:486-490.

77. Rehman, A. and Saba, T. (2011b). Document Skew

Estimation and Correction: Analysis of Techniques,

Common problems and Possible Solutions. Applied

Artificial Intelligence, 25(9): 769-787.

78. Rehman, A. and Saba, T (2012c).Analysis of advanced

image processing to clinical and preclinical decision

making with prospectus of quantitative imaging

biomarkers, Artificial intelligence Review, pp. 1-19,

doi:10.1007/s10462-012-9335-1.

79. Rehman, A. and Saba, T (2012d) ―Neural Network for

Document Image Preprocessing‖ Artificial Intelligence

Review, DOI: 10.1007/s10462-012-9337-z.

80. Rehman, A. and Saba, T. (2012e). Features extraction

for soccer video semantic analysis: current achievements

and remaining issues Artificial Intelligence Review,

DOI: 10.1007/s10462-012-9319-1.

81. Rosenbrand K, Croonenborg J, Wittenberg J (2008)

Guideline Development. In: Teije A, Miksch S, Lucas P

(eds) Computer-based Clinical Guidelines and Protocols:

A Primer and Current Trends. IOS Press, Amsterdam: 3-

21.

82. Rosenfeld R, Shiffman R (2006) Clinical practice

guidelines: a manual for developing evidence based

guidelines to facilitate performance measurement and

quality improvement. Otolaryngol. Head Neck Surg Off

J 135:1-28

83. Ruzicka M, Svatek V (2004) Mark-up based analysis of

narrative guidelines with the Stepper tool.

84. In: Kaiser K, Miksch S, Tu SW (eds) Computer-based

support for clinical guidelines and protocols: Proc Symp

Comput Guidel Protoc. IOS Press, pp 132-136.

85. Saba, T. Sulong, G. and Rehman, A. (2011a). Document

Image Analysis: Issues, Comparison of Methods and

Remaining Problems. Artificial Intelligence Review,

35(2):101-118, DOI: 10.1007/s10462-010-9186-6.

86. Seto E, Leonard K, Cafazzo J, Barnsley J, Masino C,

Ross H (2012) Developing healthcare rule based expert

systems: Case study of a heart failure telemonitoring

system. Inter J Med Inform.

doi:10.1016/j.ijmedinf.2012.03.001

87. Seyfang A, Martinez-Salvador B, Serban R, Wittenberg

J, Miksch S, Marcos M (2007)

88. Maintaining formal models of living guidelines

efficiently. In: Bellazzi R, Abu-Hanna A, Hunter J (eds)

AIME '07 Proc 11th Conf Artif Intell Med. Springer-

Verlag, Berlin, pp 441-445.

89. Shahar Y, Miksch S, Johnson P (1998) The Asgaard

project: a task-specific framework for the application

and critiquing of time-oriented clinical guidelines. Artif

Intell Med 14(1-2): 29-51.

90. Shahar, Y, Young O, Shalom E, Mayaffit A, Galperin,

M, Moskovitch R, Hessing A (2004) A framework for a

distributed, hybrid, multiple-ontology clinical-guideline

library, and automated guideline-support tools. J Med

Inform 37: 325-344.

91. Shortliffe E (1984) The adolescence of AI in medicine:

will the field come of age in the '90s? Artif Intell Med

5(2): 93-106

92. Sonnenberg F, Hagerty C (2006) Computer-Interpretable

Clinical Practice Guidelines. Methods Inf Med 45(1):

145-158.

93. Sordo M, Ogunyemi O, Boxwala A, Greenes R (2003)

Descriprion Status Update on GELLO: a Proposed

Standardized Object-Oriented Expression Language for

Clinical Decision Support. Stud Health Technol Inform

107: 164-168

94. Sutton D, Fox J (2003) The syntax and semantics of the

PROforma guideline modeling language. J Am Med

Inform Assoc 10: 433–443.

95. Straszecka E (2004) Clinical Knowledge Representation

in Terms of IF-THEN Rules and the Dempster-Shafer

Theory. In: Rutwoski L, Siekmann J, Tadeusiewicz R,

Zadeh L (eds) Artificial Intelligence and Soft Computing

- ICAISC 2004. Springer, Berlin, pp 1056- 1061.

96. Susskind R (1987) Expert Systems in Law: A

Jurisprudential Inquiry. Clarendon Press, Oxford.

Thomson P (2000) Implementing evidence-based health

care: the nurse teachers‟ role in supporting the


2658

dissemination and implementation of SIGN clinical

guidelines. Nurs Educ Today 20 (3): 207-217.

97. Thomson R, Lavender M, Madhok R (1995) How to

ensure that guidelines are effective. BMJ 311:237-242.

98. Tu S, Campbell J, Glasgow J, Nyman M, McClure R,

McClay J, et al. (2007) The SAGE Guideline Model:

Achievements and Overview. J Am Med Inform Assoc

14(5): 589–598.

99. van Gerven M, Taal B, Lucas P (2008) Dynamic

Bayesian networks as prognostic models for clinical

patient management. J Biomed Inform 41: 515-529.

100. Vetternlein T, Mandl H, Adlassnig K-P (2010) Fuzzy

Arden Syntax: A fuzzy programming language for

medicine. Artif Intell Med 49: 1-19.

101. Votruba P, Miksch S, Kosara R (2004) Facilitating

knowledge maintenance of clinical guidelines and

protocols. In: Fieschi M, Coiera E, Li Y-CC (eds) Proc

Medinfo 2004 World Congr Med Inform, AMIA. IOS

Press, San Francisco, p. 5761.

102. Wang D, Peleg M, Tu S, Boxwala A, Ogunyemi O,

Zeng Q, et al. (2004) Design and implementation of the

GLIF3 guideline execution engine. J Biomed Inform

37(5): 305-318.

103. Weidjen T, Boivin A, Burgers J, Schuneman H, Elwyn

G (2011) Clinical practice guidelines and patient

decision aids: An inevitable relationship. J Clin

Epidemiol: 1-6

11/3/2012

expert system for offline clinical guidelines and treatment · expert system for offline clinical...

Documents