KNOWLEDGE BASED CONCEPTS FOR DESIGN SUPPORT OF

AN ARTIFICIAL ACCOMMODATION SYSTEM

K. P. Scherer

Forschungszentrum Karlsruhe GmbH, Institut für Angewandte Informatik , P.O. Box 3640, 76021 Karlsruhe, Germany

Keywords: Knowledge based system, rule ba

sed design, accommodation system, frame based representation.

Abstract: When conceiving medical information and diagnosis systems, knowledge based systems are used to

diagnose failures based on specific patient data. The knowledge is evaluated based on statistical data from

the paste and the present information is derived by statistical approaches (Bayes theorem) and analogues

cases to interpret the individual patient related situation. An analogue methodical situation is that of the

conceptualisation of a new technical system, where the system components with the properties will be

configured in such a manner, that a target function is guaranteed under consideration of any constraints. In

both situations, the system (human being, technical system) has to be described in a natural language and

must be formalised. Based on these formulas logical conclusions can be drawn. Useful representations are

formalised knowledge representation methods. For logical conclusions the predicate calculus of first order is

used. For information access by both experts and users, comfortable natural language based concepts and

the employment of graphical tools are very important to manage the complex knowledge.

1 INTRODUCTION

Concerning reconstitute the vision quality for

presbyopes human eyes, a novel artificial

accommodation system (AAS) will be investigated

and developed. This is essential to guarantee life

quality at an advanced age. In the course of this

preliminary research and development vast

information will be generated in this context which

must be further processed, deleted, extended and

changed during the current process. To do this, it is

necessary to generate and dispose of software

structures for storing and processing the information

used in a consistent manner (Scherer et al., 2006).

Formal representations have to be designed such, that

the information is accessed in a natural language.

Based on this acquired expert knowledge, new

knowledge can be produced and hypotheses can be

validated or disproved. The important topic in this

proposal is the presentation of frame based concepts,

where an inference engine starts with logical

methods. Furthermore, causal relations and

interactive effects between the different components

of the accommodation system have to be regarded in

the design process.

2 BIOLOGICAL BACKGROUNDS

From the biological and medical points of view, a

mechatronic system is developed for implantation

into the capsule bag of a presbyope human being

after a cataract disease (Bergemann et al., 2006).

This mechatronic system must be self sustaining.

Consequently it needs an independent energy supply

system to reproduce the refraction power needed.

model with capsule bag.

with a capsule

system (right side)

ration. After

pty capsule bag

should be filled with the mini urised complex self

Figure 1: Human eye and a lens

Fig. 1 shows a natural eye (left side)

bag and an intraocular lens

before starting the cataract ope

extraction of the human lens, t

he em

241

P. Scherer K. (2007).

KNOWLEDGE BASED CONCEPTS FOR DESIGN SUPPORT OF AN ARTIFICIAL ACCOMMODATION SYSTEM.

In Proceedings of the Second International Conference on Software and Data Technologies - PL/DPS/KE/WsMUSE, pages 241-246

 SciTePress

nowhere existent at time.

are methods to solve the

and needs also different volumes

partial

n power, the energy

Figure 2: Different domain specific classes with relations.

installation space is limited by an upper and a lower

different power densities

ponent itself, the haptics and the

tself the

: the volume of the capsule bag in the eye

on volume of the optical component

y and the

about

”

ustaining accommodation system, which is So, the above mentioned aspects of the information

entities and their subcomponents give rise to the

following question:

Are there special conditions and is there a

special parameterisation of the mechatronical

components to reach the patient specific refraction

power desired.

The first very important requirement refers to the

volume of the different hardware components. It

decides whether a special refractive system can be

implanted with a special parameterised energy unit

of adequate power. Related former aspects,

following conditions have to be considered.

1) A patient specific quantitative volume of th

When developing such a new complex

mechatronic system, much information is produced,

which is partially wrong, uncertain and

contradictory. In spite of various constraints, the

information should be processed logically. The

management of such a complex information system

requires the use of softw

following problem:

Are there any energy supply methods for such a

self sufficient implant, taking into account the

smallness, biological compatibility and the

absolutely necessary energy to guarantee the

refraction power needed?

In this problem following information entities are

involved:

The patient or human being, undergoing the

operational intervention to insert the implant.

The type of energy supply, which has to be

selected from many possible physical, chemical and

biological components.

The refractive lens system, which has to be

implanted

depending on the different actuator principles. Such

a refractive system consists of different

subsystems (optics, haptics, actuators, sensor

components) with their own functionality. To

provide the patient with refractio

vel must be greater than a special conditioned

value.

The sensor systems, which are responsible for

the active control of accommodation, must also be

implanted into the patient specific capsule volume.

This results in a domain specific semantic network

with different class structures.

capsule bag is available for the AAS. But the

value.

2) Different physical, chemical and biological

energy supplies posses

depending on their activity principle and the degree

of efficiency.

3) The regarded refractive system consisting of

the optical com

actuator component needs space, which is described

not only by the pure volume, but also under

morphological aspects (it may be compact, cliffy,

cylindrical and so on) and affects other hardware

components.

4) Apart from to the optical component i

sensor component (measuring and control unit) has

to be installed. The following parameters are

introduced to describe the basic situation:

- Vol_P

- Vol_E: volume of the energy supply unit needed

- Vol_S: minimal volume of the sensor component

- Leistdichte_S: power density for the energy

- Vol: installati

- Leist_O: needed power for the refraction desired

3 SEMANTIC NETWORKS FOR

KNOWLEDGE

REPRESENTATION

The information embedded into the class structures

and the attributes, are correlated to formalise the

knowledge about patients, energy suppl

refractive system (requirements see chapter 2). T

main information entities (patient, energy supply,

refractive system) are regarded as classes wit

variable objects (Fig. 3). The additional slots

these classes are the two relations “has properties

and “consists of”. The properties are inherited fro

capsule

bag

refr.

system

sensor

comp

causal

relation

optical

comp

energy

supply

ICSOFT 2007 - International Conference on Software and Data Technologies

242

sup r classes along the relation path “is a”. For th

inf mation system a frame based approach i

tructed for the classes themselves, wh

nded further to so called conc

e e

or s

cons ich will be

exte epts (concept =

frame with attributes including restrictions).

Figure 4: Semantic network of class and object relations.

The class of the refractive optical systems is

subdivided into five classes, each distinguishing by

cons s system. Other

principles and optical systems are ignored in this

ne, that is based on

ibration generators”. The attributes of interest are

e power density and the volume needed.

4 FRAME BASED KNOWLEDGE

REPRESENTATION

The modelling of the natural information about

mentioned semantic topics means the development

of structures, which are formulised in class-subclass-

element-relations. Along the relation “is a” between

classes and subclasses and elements an inheritance

mechanism is available. For the attribute slots,

constraints can be formulated. These frame based

approaches form a well defined hierarchical tree

structure for the new information system and are a

prerequisite for starting logical conclusions.

4.1 Class Structure of the Refractive

Systems

Figure 3: Class components with relations and attributes

For the overall correlated information a rule

based approach is applied with the known Boolean

operators of the predicate calculus of first order. The

evaluation of the attributes in the classes patient

refractive system and energy supply facilities form

the symptom tree, otherwise any information about

fit accuracy of the participating components in the

capsule bag compose the diagnostic tree. The

relations between this hierarchical organised

knowledge are performed by rules with functions

and interpretations. (Fig. 4).

the actuator principle and the physical principle. In

the following application only one class will be

idered, namely the elastic len

stage of knowledge engineering. The elastic lens

itself is studied and the aspect “consists of”. Each of

the resulting two classes of optical device (optics)

and sensor device (sensors), has “installation space”

(volume) as an attribute. Along the relation “consists

of” no inheritance is available.

Figure 5: Class hierarchy of refractive optics (excerpt).

4.2 Class Structure of Energy Supply

For the development of the AAS different energy

supply units are analysed. A representation of four

energy supply classes is needed. The attributes of

these classes have both numerical and also linguistic

values. Here it is focussed on a single energy supply

class, namely the mechanical o

“v

energy supply

power density

install. space

props

props.

patient

inst.space

props

optics

install.space

refract. system

sensors

haptics

morphology

consists of

props

energy

props

refract

optics

fluidic

len

elastic

lens

electrow

etting

Alvarez

lens

triple

optics

optics sensors

symptoms

diagnostic

refractive system

- elast. lens

Alvarez

patient

energy supply

- mechanical

chemical

prob.

density

fit

no fit

logical inference

fit

accuracy

inst.space

energy

vol

rules, functions

inter

retations

KNOWLEDGE BASED CONCEPTS FOR DESIGN SUPPORT OF AN ARTIFICIAL ACCOMMODATION SYSTEM

243

Figure 6: Class hierarchy of energy supply units (excerpt).

4.3 Class Structure of Patients

A third important class in connection with the fitting

accuracy in the capsule bag is the set of all patients.

At the moment, the volume value is

the only

umerical feature that describes the installation

The rule based structures refer to the domain

d ey reflect

the causal relationships between the diagnostic part

space of intere

st. Other also important features like

the biological compatibility and other medical

constraints are added in time.

Figure 7: Class patient with the attribute installation space

(volume).

5 RULE BASED STRUCTURES

depen ent part of the knowledge base. Th

of fitting accuracy in the human capsule bag and the

prerequisites, formulised by the frame based classes

patient, energy supply and refractive system. A

simple rule

is following:

ithout Boolean operators

ctive function in knowledge processing

t in

Figure 8: Simple rule w

The rule only describes the dependence status and

has no a

con rary to a rule interpreter. This will be realised

the i ference engine.

energy

mechanical

electro

netical

chemical

fluidics

vibration

As an extension of the former rule a complex rule

consists of atomic preconditions associated by

operators of predicate calculus of first order.

Figure 9: Complex rule with AND conditions.

In the next stage, structured rules are complex rules

with exception conditions being specified.

5.1 Rule Extension with

Interpretations

S etimes, the numerical values of attributes are

n for clusions. Before

fi evaluation of the

at pe reason with the

co ation

sum of all

p has ated before a

co n can be drawn with respect to the fitting

ac ower density

and the volume results in the power, which is one of

the two AND conditions for the conclusion process.

ot input directly logical con

nal conclusion, an algorithmic

tributes must be rformed to

ndensed inform .

As shown in figure 10 first the

arameter values to be calcul

nclusio

curacy. Also the product of energy p

Figure 10: Rule extension with interpretations

6 ARCHITECTURE OF THE

KNOWLEDGE BASED SYSTEM

A knowledge based system is mainly a computer

application, which performs a special task, which

would be performed usually by human experts.

These systems are part of the general category of

power

densit

volume

patient

inst. space

e le bag

AND Vol_O + Vol_S + Vol_E < Vol_P

AND Leistdichte_S x Vol_E > Leist_O

then

nergy supply fits into capsu

Vol_O < 107 mm

then

AND Vol_O < 107 mm

AND Vol_S < 20 mm

AND Leist_O > 15 mW

ND Vol_P > 320 mm

eistd mW/mm

l_E 0 mm

AND L ichte_S > 80

Vo < 19

AND

then

energy supply fits into capsule bag

ICSOFT 2007 - International Conference on Software and Data Technologies

244

artificial intelligence applications and capture an

expert’s decision making knowledge, such that it can

be disseminated to others. The knowledge based

systems differ from conventional programs by

performing tasks using decision making logics under

constraints and other conditions. The kernel of such

a system exists of the two main components the

knowledge base (subdivided into the two parts “fact

base” ine,

which f the

knowledge base to draw conclusions concerning a

c ortable

acqu

A special feature of the knowledge based

fact base.

ge domain. Depending on a

ate (given by formulised “if conditions” as

p tems are derived. In this

case only categorical knowledge is used. The “if

c ed by

p operators. This condition is sufficient for

t ural

(

he active part of the

and “rule base”) and the inference eng

uses the parameterised facts and rules o

spe ial given problem. Additionally a comf

exp anation component and a consistent knowledge

isition tool complete the architectural structure.

architecture is the strongly logical division into the

passive knowledge base and the active strategic part

of making logical conclusions (Puppe et al., 2001).

Figure 11: Architecture of the knowledge based system

6.1 Fact Base

The current situation is represented by the facts, i.e.

the parameterised attributes of the classes and

objects. The different facts are stored in the fact

base. In the presented special application the

situation is as follows:

Figure 12: Situation in

6.2 Rule Base

The different representations concern the rule base,

describing the relations among the knowledge

entities in the knowled

remises) new knowledge i

onditions” contain atomic formulas connect

redicate

he declarative (hypotheses) and the proced

actions) conclusion parts.

Figure 13: Domain specific rule base.

6.3 Inference Engine

he inference engine is t

domain specific knowledge

set of causally connected information units

predicates

formal representation:

B1 op B2 op…satisfied

conditions

nowledge based kernel to process the given

information and obtain new results. The conclusion

result from the logical predicate calculus of first

order. The definition is as follows:

An inference rule is an advice of how to generate

a new formula in a logically correct manner by

combining two (or any) given logical formulas. An

important inference rule used in this application i

the so called “modus ponens”, given by the

following method.

A (fact in fact base)

A >B

(rule in rule base)

B (new fact generated as conclusion

written into the fact base)

In this c tself are

ty. A

nditions concerning

ontext, A, B and the implication i

deterministic and without loss of certain

contains all parameterised preco

the patient’s capsule bag, the refractive system with

components and the energy supply unit.

Then

actions A1, A2 ….

hypotheses D1, D2…

procedural

declarativ

fact base

rule base

domain s

ecific

inference

knowledge base: intermediate and final results

(conclusions, newly generated knowledge)

acquisition

explanation

Case specific knowledge

In this situation the following parameters are

given:

A volume of the space for installing the optics

A special parameterised sensor component

A patient specific capsule bag

A special energy supply unit with attributes

KNOWLEDGE BASED CONCEPTS FOR DESIGN SUPPORT OF AN ARTIFICIAL ACCOMMODATION SYSTEM

245

B contains information about the fitting accuracy of

the artificial accommodation system

into the capsule

ented so far.

ion

The human bas nguage)

ransformed into formal structures of the

tion step.

onsistence means a validation of values in a

The expl ponsible for

providing s as to how the

reasoning tial to

understan obtain new

ideas for f sses. In this

way, con onclusion and the

developed knowledge based system is enhan d.

solution paths (numerous conditions and constraints)

e user specific results have to be analysed by

components.

Together with the inference engine they form the

r powerful

navigation through the complex knowledge system.

comfortable decision making processes including

mic based system has

to be extended to a knowledge based system for

h, U., Bretthauer, G., Guthoff,

R., 2006, Artificial Accommodation System – a new

approach to restore the accommodative ability of the

ld Congress on Medical Physics

eering. Seoul.

bag. The instantaneous state of diagnostics is a

binary valued statement “the system can be

implanted” (based on value conditions) or “system

cannot be implanted”.

A weak uncertainty function of accuracy exists but

has not been implem

6.4 Knowledge Acquisit

ed knowledge (natural la

must be t

computer software components. The knowledge

entities are inputted into the class-objects-attribute

variables (see also frame based structures).

Knowledge acquisition can be enhanced by

consistence checking during the acquisi

predefined numerical range or linguistic values

within a predefined term set or the number of

allowed values from the whole acquisition set.

Figure 14: Frame extension to concepts (with facets).

This indirect constraint representation prevents

formally correct conclusions from being drawn on

the basis of data that are not allowed. Extension of

the frames to the so called concepts is a useful

representation of the knowledge schemes. An

example is shown in Fig. 14.

.5 Explanation Component 6

anation component is res

the user with explanation

process was performed. It is essen

d the reasoning process and

urther advanced solution proce

fidence into the c

Due to the probably very complex and wid

backward chaining processes. This requires a

comfortable capacity to interact with the system

through text and graphics.

The explanation part together with the knowledge

acquisition component are the dialogue

monitoring process, that is responsible fo

7 CONCLUSION

The benefit and necessity of knowledge based

structures in the research and development of a

novel complex artificial accommodation syst

(ASS) are outlined. The acquired knowledge must

be managed using refinement processes, because

absolute information is lacking in the instantaneous

state.

Original rules developed may be rewritten later

and redefined. The comp

lex knowledge is more

circular than linear. Furthermore the knowledge

always is only partly correct and not complete and

has to be redefined gradually. In this meaning

comfortable formulised structures and refinement

mechanisms must be developed as well as

their explanation.

Hence, the classical algorith

developing the new complex artificial

accommodation system.

REFERENCES

Scherer, K.P., Guth, H., Stiller, P., 2006, Computational

Biomechanics and Knowledge based Structuring of

Human Eye Components, International Congress on

Applied Modelling and Simulation, Rhodes, Greece,

June 26-28, 2006 ISBN 0-88986-561-2.

Scherer, K.P., Guth, H., Stiller, P., 2006, Solid and mesh

modelling concepts for natural and artificial eye

components, International Congress on Modelling,

Instrumentation and Control, Lanzarote, Spain, Feb.

5-2, 2006 ISBN 088986-551-5.

Puppe, F., Reinhardt, B., 2001, Multimediale wissens- und

fallbasierte Trainings- und Informationssysteme in der

Medizin, BMFT-Verbundprojekt, Projektträger DLR,

Förderkennz: 01EI9603/1,2001.

Bergemann, M., Gengenbac

human eye. In: Wor

and Biomedical Engin

AND 100 < Vol_O < 110

AND 15 < Vol_S < 25

AND Leist_O > 15

AND 280 < Vol_P < 3

AND Leistdichte_S in mW

AND Vol_E in mm

ICSOFT 2007 - International Conference on Software and Data Technologies

246