PROCESS AND E-SERVICE CUSTOMIZATION

For Coordination in Healthcare Networks

Günter Schicker, Carolin Kaiser and Freimut Bodendorf

Department of Information Systems, University of Erlangen-Nuremberg, Lange Gasse 20, 90403 Nuremberg, Germany

Keywords: Healthcare networks, processes, case based reasoning, coordination, customization, e-services.

Abstract: Coordination in healthcare networks becomes increasingly important. A process-oriented coordination

approach is introduced which enhances integrated care scenarios by an IT-driven coordination of

interorganizational treatment processes – the concept of process-based e-service logistics. The allocation of

e-services is based on a model describing services and coordination tasks between roles in a healthcare

network. The underlying system’s architecture is presented which implements process-based e-service

logistics by designing and executing individual treatment processes, identifying coordination tasks between

network actors and dynamically allocating e-services. A solution for automated individualization of

processes and e-services based on Case Based Reasoning (CBR) technology is discussed.

1 HEALTHCARE NETWORKING

The healthcare industry is an important economic

sector in Germany causing annual expenses of about

230 billion euros and employing more than 4.2

million people. Starting point of the research project

was an empirical study addressing German and

Swiss ambulant healthcare networks (healthcare net-

work managers as well as physicians) to evaluate the

maturity of healthcare network organizations regar-

ding strategy, processes and information technology

(Schicker et al., 2006a). The empirical study reveals

that 81 percent of the respondents expect that net-

working in the healthcare industry will increase in

the next three to five years. Moreover, 88 percent of

the survey participants agree that the demand for

coordination and IT-support in healthcare networks

is going to rise in the future (Schicker et al., 2006a:

21). An important driving force of networking is the

concept of integrated care which is often associated

with the following instruments:

 Intersectorial cooperation (Ramming, 2004: 147;

Mühlbacher, 2002: 65)

 Financial responsibility (e.g. capitation)

 Coordination and control of medical treatment pro-

cesses (Mühlbacher, 2002: 66)

 Information integration (Ramming, 2004: 147).

Whereas many research projects deal with the in-

tegration of health data this project concentrates on

the coordination and control of interorganizational

processes within healthcare networks by providing

patients and suppliers with a customized set of elec-

tronic services. To analyze the requirements the re-

search team cooperates with the healthcare network

“Qualitäts- und Effizienzgemeinschaft Nürnberg-

Nord” which is organized as a gatekeeper system.

The integrated care contract spans ambulant, clinical

and home care service providers and is financed by a

full capitation model (Wambach et al., 2005: 13).

2 PROCESS-ORIENTED

COORDINATION

Process-oriented coordination is seen as one

important way to enable integrated care scenarios, to

enhance patient satisfaction and to reduce costs of

treatment processes (Güntert, 2004: 100ff).

Schmalenbach argues for process-oriented control

especially to cope with the problem of managing

multiple interfaces (Schmalenbach, 1908/09; 211f.).

Based on that belief the research project focuses on

the treatment process from a cooperative view

regarding the patient’s way throughout the whole

healthcare network.

2.1 Process Characteristics and

Requirements

Process-oriented coordination faces important

challenges to cope with. Therefore the characteris-

161

Schicker G., Kaiser C. and Bodendorf F. (2008).

PROCESS AND E-SERVICE CUSTOMIZATION - For Coordination in Healthcare Networks.

In Proceedings of the First International Conference on Health Informatics, pages 161-166

 SciTePress

tics of healthcare network processes and their requi-

rements will be discussed next. To transfer the gene-

ral tasks and principles of coordination to the health-

care domain, it has been analyzed which processes

(e.g. management, medical treatment and support

processes) and coordination tasks exist within

healthcare networks (Schicker, 2006b: 39).

2.1.1 Process Characteristics

Table 1 depicts characteristics of interorganizational

treatment processes that need to be considered when

requirements for the IT-support in healthcare net-

works are defined (Schicker et al., 2006b: 39).

Especially the uniqueness of one patient´s treatment

process (Müller-Mundt, 2001: 95) and the high

degree of volatility are key characteristics which

have to be considered when supporting healthcare

network processes by information technology. Du-

ring their execution processes have to be modified,

detailed and customized to the patient´s needs de-

pending on his individual treatment context (cyber-

netic model of medical treatment) (Prokosch, 2007).

Table 1: Characteristics of treatment processes.

• unique

• requiring intensive

coordination and

information

• cross-sectorial

• important and risky

• stepwise changing and

volatile, requiring in-

depth knowledge

• involving numerous

participants

• long-lasting and complex

2.1.2 Requirements

The individual characteristics of each patient, the

high degree of volatility during the real-time execu-

tion of the process instance are crucial challenges

when supporting modelling, adaptation and execu-

tion of individual processes by information techno-

logy (Purucker et al., 2007).

For that reason process models should be ad-

apted to the individual needs of the patients (Haas,

2005: 553ff). To avoid modelling efforts a new auto-

mated process design is needed.

Moreover, members of the healthcare network

(especially gatekeepers) should be able to modify

the individual process model easily as soon as new

information about the state of treatment or illness

exists (Schwarz et al., 2001: 10; Remus, 2002: 115).

Finally, e-services (e.g. information and application

services) must be linked to individual process activi-

ties in a flexible way.

3 PROCESS-BASED E-SERVICE

LOGISTICS

3.1 Basic Principles

The concept of process-based e-service logistics

(PEL) aims to support the coordination of healthcare

network processes by providing patients and health-

care suppliers with a customized set of electronic

services. E-services are software components which

encapsulate functions (e.g. logic or data centric ser-

vices) in a coarse-grained manner, e.g. using web

services as technical representation (Krafzig et al.,

2004: 70ff). The e-service requirements regarding

information and coordination in networks are deri-

ved from customized process models. They result in

a process-based e-service logistics model executed

by a process management platform supporting the

coordination of individual treatment processes by

providing network participants with e-services.

3.2 Architecture Overview

The proof of the concept described above is being

implemented and called Individual Value Web Sys-

tem (IVWS). IVWS is a process platform supporting

configuration, execution and control of processes

across healthcare networks. The system architecture

follows a four-tiered approach (see figure 1).

3.2.1 Presentation Layer

Application front ends initiate and control all activi-

ties of the IVWS. Typical application front ends are

graphical user interfaces enabling direct interaction

of users with the system (WebParts implemented

with C# using the Sharepoint Portal Server). Within

the research project role-specific process portals for

patients, service providers and network managers are

implemented (Schicker et al., 2005: 7ff).

3.2.2 Customization and Flow Control

Layer

This layer consists of three components including

features for orchestration and execution of web ser-

vices.

 Service bus: This component ensures the exe-

cution of web service orchestrations and pro-

vides functionalities of a service bus: connec-

tivity, integration and communication ser-

vices, process orchestration and execution

(Krafzig et al., 2004: 65).

 Process and e-service customization: An im-

portant issue of PEL is how to adapt and

HEALTHINF 2008 - International Conference on Health Informatics

162

model one patient’s treatment process and the

underlying e-services needed for supporting

the process (for details see section 4).

 Meta orchestration server (MOS): MOS enab-

les the configuration, execution and monito-

ring of treatment processes. MOS represents a

process engine which acts as a broker between

the front ends and the e-services invoked du-

ring process execution. It also initiates the

execution of web service orchestrations by the

BizTalk Server and integrates process propo-

sals created by the customization component

for process individualization (for details see

Purucker, 2007).

3.2.3 Application Layer

This layer contains all e-services needed for exe-

cution and support of treatment processes. E-ser-

vices of a SOA are software components encapsula-

ting business functions in a coarse-grained manner.

An e-service consists of a service contract, a service

interface and an implementation. The imple-

mentation is the physical representation of the re-

quired business logic and the relevant data (pro-

grams, configuration data and database) (Krafzig et

al., 2004). E-services include IVWS-specific ser-

vices (e.g. the customization component described

below) as well as third party web services and

adapters for the integration of existing information

systems.

3.2.4 Data Layer

A MS SQL-based process and e-service repository

provides features for searching and applying process

models and e-services, e.g., physical location, ser-

vice provider, charge fee, technical constraints, se-

curity data and service level agreements.

4 PROCESS AND E-SERVICE

CUSTOMIZATION

4.1 Traditional Methods

There are many approaches which enable the

customization of processes in general and which can

be applied for the customization of treatment pro-

cesses and e-services. Following the classic ap-

proach, each treatment process model and each e-

service is manually developed from scratch for each

context with respect to guidelines (Lang, 1997: 29).

The steady process of creating customized treatment

processes and e-services is very time-consuming and

inhibits the reuse of experiences. To overcome this

disadvantage new concepts have been developed.

The process modelling approaches based on refe-

rence models (vom Brocke, 2003: 31ff), process ske-

letons (Remme, 1997:114ff) and process modules

(Lang 1997: 4ff) enable the reuse of treatment pro-

cess knowledge. The approaches for creating e-ser-

vices by using design patterns, frameworks and li-

braries allow the reuse of e-service knowledge.

However, these concepts require a time-consuming

search and adaptation to the current context.

The disadvantage of existing approaches (e.g

Wargitsch et al., 1997: 3ff; Rupprecht, 2002: 67ff) is

that only one of the two steps (search and adapta-

tion) is automated whereas the other step must be

executed manually. An approach is needed which

enables an automatic search and adaptation

(Schicker et al., 2007). Here, such a system based on

Case Based Reasoning (CBR) is introduced.

Service

Provider Portal

Data

MS SQL Server

Application

E-Services of Business Objects

ADO.NET

E-Services (third party)

Web

Services

Web

Services

AdapterAdapter

Service

Provider

Performance

Cockpit

Network

Manager

Treatment

Process Portal

Patient

Treatment

Process Portal

Patient

Presentation

Customization

and Flow

Control

Service Bus

MS Biztalk Server

Meta-Orchestration-Server (MOS)

Process and

E-Service

Customization

(CBR)

User Login Process State Configuratio n

Administration

Broker

Figure 1: IVWS architecture.

PROCESS AND E-SERVICE CUSTOMIZATION - For Coordination in Healthcare Networks

163

4.2 Case Based Reasoning

4.2.1 Introduction

CBR is a problem solving paradigm of Artificial In-

telligence. It solves new problems based on past ex-

periences saved in form of cases in the case base.

Each case (patient record) consists of a problem de-

scription (patient context) and a solution (treatment

process and its e-services). In order to propose a

treatment process and e-services for a given patient,

CBR searches its case base for the patient record

which is most similar to the given patient record and

adapts its treatment process with e-services to fit the

given case.

Besides fulfilling the requirements of automated

search and adaptation, CBR offers several advanta-

ges (Nilsson and Sollenborn 2004: 178). First of all,

it resembles the physicians’ cognitive process of re-

calling former patients and reusing past experiences.

This resemblance does not only prove the force of

the approach but also leads to high user acceptance.

Furthermore, the collection of patient records can

easily be integrated in a CBR system as a case base.

Moreover, the reuse of patient records provides an

efficient reasoning mechanism which does not

require solving each problem from scratch.

The functionality of a CBR system can be divi-

ded into four main phases which form the CBR

cycle: retrieve, reuse, revise, retain phase (Aamodt

and Plaza 1994: 46). The CBR cycle for customizing

treatment processes and e-services is illustrated in

figure 2 and described in the following sections.

4.2.2 Representation of Patient Records

Each patient record (case) contains the patient con-

text (problem) as well as its treatment process and e-

services (solution). While the patient context spe-

cifies the patient and his disease pattern, the treat-

ment process and its supporting e-services charac-

terize the therapy of the patient.

Attribute-value-vectors are used to represent the

patient context. According to this type of representa-

tion, the patient context is specified by a vector of

attribute values. In order to reflect the characteristics

of different diseases, the set of attributes has to be

defined for each disease separately.

Treatment processes and e-services can consist

of several elements which are represented in an

object-oriented manner. They are specified by a set

of instances of classes (e.g. treatment process,

service, coordination task, patient task and e-ser-

vice).

4.2.3 Retrieve Phase

In the retrieve phase a health care provider passes

the attribute values of the patient context to the sys-

tem and requests a treatment process. In order to ful-

fil this request, the system searches for the patient

record on the case base whose patient context is

most similar to the context of the given patient. The

search for a similar case requires the definition of

similarity measures and search algorithms.

The similarity calculation of two patient contexts

is based on local and global similarity measures

(Stahl, 2003: 50ff). Local similarity measures deter-

mine the similarity between one attribute-value of

the query patient record and the patient record in the

case base, whereas global similarity measures calcu-

late the aggregated similarity of all attribute-values.

Different types of local similarity measures are used

depending on the type of the attribute.

Figure 2: CBR system for customizing treatment processes and e-services.

HEALTHINF 2008 - International Conference on Health Informatics

164

For determining the similarity of nominal attributes,

similarity tables are appropriate. They assign a

similarity value to each combination of attribute

values of the query patient context and the patient

record in the case base. In order to calculate the

similarity of metric attributes threshold-based,

linear, exponential and sigmoide functions can be

applied.

A weighted sum of the local similarities is cho-

sen as global similarity measure. The weights reflect

the relevance of the attributes. In order to reduce the

high effort for the manual definition of similarity

measurements, a learning algorithm for simplifying

and optimizing the global similarity measurement

has been implemented.

For searching the most similar patient records,

sequential search and knowledge-poor indexing ba-

sed on an extended k-d-tree (Wess, 1995: 163ff,

209ff) are provided (for details see Kaiser, 2008).

4.2.4 Reuse Phase

The reuse phase receives as input the record of the

most similar patient which was found during the re-

trieve phase and aims at adapting it to fit the new pa-

tient record. The adaptation for the customizing of

treatment processes consists of four successive

steps: copy adaptation or compositional adaptation,

substitutional adaptation, structural adaptation and

consistency assurance.

The first adaptation step takes different turns de-

pending on the phase of the treatment process. If a

new treatment process should be created the treat-

ment process of the similar patient record can be co-

pied. However, if an existing treatment process must

be extended, elements of the similar patient record

are added. The substitutional adaptation replaces the

attribute values of treatment elements and e-service

elements adopted from the similar patient record

according to rules. The structural adaptation modi-

fies the structure of the requested treatment process

and its e-services by deleting, adding and rearran-

ging elements. It is realized by an additional adapta-

tion-based CBR system (see figure 2). The adapta-

tion cases of the adaptation-based CBR system des-

cribe which modification actions (solution) should

be taken when certain differences and mutualities of

the patient context (problem) between query and re-

cord in the case base occur (for details see Kaiser,

2008). The consistency assurance aims at improving

the consistency of the created treatment process and

its e-services.

4.2.5 Revise Phase

During the revise phase, the execution of some treat-

ment steps and e-services takes place. In order to

support the execution, the treatment process and its

e-services are transformed into an XML represen-

tation and passed on to the meta-orchestration server

for further processing.

4.2.6 Retain Phase

The retain phase is the last phase of the CBR cycle

and belongs to the case base maintenance. The aim

of the case base maintenance is to detect environ-

mental changes which could decrease the quality and

efficiency of the case base and execute counteractive

measures. Besides the retain phase, the restore phase

and the review phase are also part of the case base

maintenance (Roth-Berghofer 2002: 55ff). The re-

tain step uses intra-case quality measures to decide

on the insertion of a patient record and is called each

time a patient record has passed the revise phase.

The review phase and restore phase are executed

periodically. Hereby all patient records in the case

base are checked with the aid of inter-case quality

measures and redundancy measures and added or

deleted accordingly.

5 CONCLUSIONS

The importance of coordination and IT-support in

healthcare networks is increasing steadily. In order

to enhance the quality and efficiency of interorgani-

zational treatment processes, an IT-supported pro-

cess-oriented coordination approach for healthcare

networks has been realized. The approach supports

the configuration, execution and control of processes

and supporting e-services in healthcare networks.

This paper focuses on the configuration of treat-

ment processes and e-services. Uniqueness and vo-

latility are key characteristics of treatment processes.

Treatment processes and e-services have to be

configured whenever new information on the status

of the patient becomes available. In order to reduce

the manual effort for searching and adapting treat-

ment processes and e-services an automated system

based on CBR has been developed.

The definition and update of the patient context

attributes, treatment process steps and e-services is

time-consuming. However, it enables to formalize

and expatiate on medical treatment knowledge. The

approach offers basic functionality and can be

extended by further methods such as the prediction

of unknown attribute values of the patient context.

PROCESS AND E-SERVICE CUSTOMIZATION - For Coordination in Healthcare Networks

165

REFERENCES

Aamodt, A., Plaza, E., 1994. Case Based Reasoning:

Foundational Issues, Methodological Variations and

System Approaches. AI Communications 7 (1), 39-59.

Güntert, B. J., 2004. Integration und Kooperation im Ge-

sundheitswesen – ein Plädoyer für patientenorientierte

Strukturen und Verhaltensweisen. In: Busse R. et al.

(Ed.): Gesundheitswesen – vom Heute ins Morgen.

WuV Universitätsverlag, Wien, 100-109.

Haas, P., 2005. Medizinische Informationssysteme und

Elektronische Krankenakten. Springer, Berlin.

Kaiser, C., 2008. Case based Reasoning for Customizing

Treatment Processes. To appear in Lazakidou, A. A.,

Siassiakos, K. M.. Handbook of Research on Distribu-

ted Medical Informatics and E-Health. Idea.

Mühlbacher, A., 2002. Integrierte Versorgung: Manage-

ment und Organisation. Hans Huber, Bern.

Müller-Mundt, G., 2001. Patientenedukation zur Unter-

stützung des Selbstmanagements. In: Hurrelmann, K.,

Leppin, A. (Ed.): Moderne Gesundheitskommuni-

kation: vom Aufklärungsgespräch zu E-Health. Hans

Huber, Bern, 94-106.

Krafzig, D., Banke, K., Slama, D., 2004. Enterprise SOA,

Service-Oriented Architecture Best Practices. Prentice

Hall PTR, Maryland.

Lang, K., 1997: Gestaltung von Geschäftsprozessen mit

Referenzbausteinen. Deutscher Universitätsverlag,

Wiesbaden.

Prokosch, H. U., 2007. Kybernetisches Modell ärztlichen

Handelns. In: Vorlesungsunterlagen zur Einführung in

die Medizinische Informatik für Informatik-Neben-

fachstudierende. Retrieved February 18, 2007, from

http://www.imi.med.uni-erlangen.de/lehre/ws

0506/medinfgrund_06.pdf.

Purucker, J., Schicker, G., Affes, F., Bodendorf, F., 2007.

Meta-Orchestration of E-Services for Process-oriented

Coordination in Healthcare Networks. In: Lecture

Notes in Informatics (LNI), Gesellschaft für Informatik

(GI). Köllen Verlag, Bonn.

Ramming, J., 2004. Integrierte Gesundheitsversorgung. In:

Jähn, K., Nagel, E. (Ed.): e-Health. Springer,

Heidelberg, 147-151.

Remme, M., 1997. Konstruktion von Geschäftsprozessen –

Ein modellgestützter Ansatz durch Montage gene-

tischer Prozesspartikel. Gabler Verlag, Wiesbaden.

Remus, U., 2002. Integrierte Prozess- und Kommunika-

tionsmodellierung zur Verbesserung von wissensinten-

siven Geschäftsprozessen. In: Abecker, A et al. (Ed.):

Geschäftsprozessorientiertes Wissensmanagement.

Springer, Berlin, 91-122.

Roth-Berghofer, T., 2002. Knowledge Maintenance of

Case Based Reasoning Systems - The SIAM Metho-

dology. University of Kaiserslautern, Dissertation.

Rupprecht, C., 2002. Ein Konzept zur projektspezifischen

Individualisierung von Prozessmodellen. Universität

Karlsruhe, Dissertation.

Schicker, G., Bodendorf, F., 2005. Portalunterstützte

Behandlungspfade in Gesundheitsnetzen: Prozesse -

Architektur - dynamische Navigation. In: Cremers, A.

B. et al. (Ed.): Lecture Notes in Informatics (LNI),

Gesellschaft für Informatik (GI). Köllen, Bonn, 7-11.

Schicker, G., Bodendorf, F., 2006. Process-based E-

Service-Logistics for Healthcare Networks. In:

Stormer H., Meier A., Schumacher M. (Ed.): Lecture

Notes in Informatics (LNI) - Proceedings. Gesellschaft

für Informatik (GI). Köllen, Bonn, 37-46.

Schicker, G., Kaiser, C., Bodendorf, F., 2007. Indivi-

dualisierung von Prozessen und E-Services mithilfe

von Case Based Reasoning. In: Oberweis, A. et al.

(Ed.): eOrganisation: Service-, Prozess-, Market-

Engineering/Band 1. Universitätsverlag Karlsruhe,

Karlsruhe, 713-730.

Schicker, G., Kohlbauer, O., Bodendorf, F., 2006. Praxis-

netz-Studie 2006 - Status Quo, Trends & Herausfor-

derungen. Arbeitspapier Wirtschaftsinformatik II Nr.

01/2006, Universität Erlangen-Nürnberg. Nürnberg.

Schmalenbach, E., 1908/09: Über Verrechnungspreise. In:

Zeitschrift für handelswissenschaftliche Forschung 3,

165-185.

Schwarz, S., Abecker, A., Maus H., Sintek, M., 2001. An-

forderungen an die Workflow-Unterstützung für

wissensintensive Geschäftsprozesse. Retrieved Octo-

ber 29, 2006, from http://sunsite.informatik.rwth-

aachen.de/Publications/CEUR-WS/Vol-

37/Schwarz.pdf.

Stahl, A., 2003. Learning of Knowledge-Intensive

Similarity-Measures in Case Based Reasoning.

Universität Kaiserslautern, Dissertation.

Nilsson, M., Sollenborn, M., 2004. Advancements and

Trends in Medical Case-Based Reasoning: An Over-

view of Systems and System Development. In Procee-

dings of the FLAIRS Conference, AAAI Press, 178ff.

Vom Brocke, J., 2003. Referenzmodellierung – Gestaltung

und Verteilung von Konstruktionsprozessen. In

Becker et al. (Eds.). Advances in Information Systems

and Management Science. Logos, Berlin.

Wambach, V., Lindenthal, J., Frommelt, M., 2005. Inte-

grierte Versorgung - Zukunftssicherung für nieder-

gelassene Ärzte. Ecomed, Landsberg, 2005.

Wargitsch, C., Wewers, T., Theisinger, F., 1997: Work-

Brain: Merging Organizational Memory and Work-

flow Management Systems. In Workshop ‚Knowledge

Based Systems for Knowledge Management in Enter-

prises’ im Rahmen der 21. KI-Jahrestagung. Freiburg.

Wess, S., 1995: Fallbasiertes Problemlösen in wissensba-

sierten Systemen zur Entscheidungsunterstützung und

Diagnostik. Infix Verlag, Sankt Augustin.

HEALTHINF 2008 - International Conference on Health Informatics

166