Work partial granted by FCT (POCTI/0753/2004) through CINTESIS n

753/2004 FCT, by KCrypt (POSC/EIA/60819/2004) and LIACC through

the Programa de Financiamento Plurianual, FCT and Programa POS

A FIRST APPROACH FOR A REGIONAL WIDE VEPR

Pedro Vieira-Marques

Informatics Department, Faculty of Medicine,University of Porto

CINTESIS, Faculty of Medicine, University of Porto, Portugal

Arthur Cunha, Luís Antunes

LIACC, Faculty of Science, University of Porto, Portugal

Ricardo Cruz-Correia, Altamiro Costa-Pereira

Department of Biostatistics and Medical Informatics, Faculty of Medicine University of Porto

CINTESIS, Faculty of Medicine, University of Porto, Portugal

Keywords: Electronic Health Records; Agents and Cooperative Systems; Integration and Interoperability.

Abstract: Patients visit multiple health institutions and leave a trail of information scattered around hospitals,

healthcare centres and laboratories. Information availability is of major importance in healthcare delivery.

Most of the Electronic Patient Record systems are unarticulated and usually address only the specificities of

a single medical specialty. Virtual Electronic Patient Records such as MAID (Multi Agent system for the

Integration of Data) system provide for the necessary means for intra-institutions departmental information

integration. In this paper is presented a mobile agent based extension to the agent based MAID system in

order to enable inter-institution patient data integration. This system was designed as a MAID extension

with additional patient data integration features. In order to accomplish this, modules for external data

discovery and collection where developed using mobile agents. Data collection activities are trigged by

scheduled clinical events. The system is intended to enhance an existing institutional system taking it

beyond the institutional barrier providing health professionals with a more complete patient clinical history.

1 INTRODUCTION

Through the years Electronic Patient Record (EPR)

systems have been developed in order to provide

physicians with structured and helpful information.

However most of these systems are unarticulated

and usually address only the specificities of a single

medical specialty. Integration of healthcare

Information Systems (IS) is essential to support

shared care in hospitals, to provide proper care to

mobile individuals and to make regional healthcare

systems more efficient.

Healthcare is recognised as one of the most

important areas for applications and services

integration.

However, to integrate clinical IS in a way that

will improve communication and data use for

healthcare delivery, research and management, many

different issues must be addressed (Berg 2001;

Littlejohns, Wyatt et al. 2003). Consistently

combining data from heterogeneous sources takes a

great deal of effort because the individual feeder

systems usually differ in several aspects, such as

functionality, presentation, terminology, data

representation and semantics (Lenz and Kuhn 2002).

Successful integration seeks ubiquity – data

stored at one place should be available elsewhere;

consistency and integrity – different data items

stored within the system should not be contradictory

and integrity constraints should be maintained;

synchronization – in particular, concurrent data

access should not result in inconsistent data; single

system image – the user should be able to interact

with the system via a singe standard user interface;

and finally transparency – the user should not be

aware of heterogeneous system components.

215

Vieira-Marques P., Cunha A., Antunes L., Cruz-Correia R. and Costa-Pereira A. (2008).

A FIRST APPROACH FOR A REGIONAL WIDE VEPR.

In Proceedings of the First International Conference on Health Informatics, pages 215-218

 SciTePress

Multiagent systems can successfully solve

problems where the required knowledge is

physically distributed in several places (for example,

they can gather patient data from different medical

institutions or discover distributed e-health services)

(Moreno 2006). Previously, we have developed

MAID (Multiagent System for Integration of Data)

(Cruz-Correia, Vieira-Marques et al. 2005), an

agent-based system that provides access to patient

data that’s scattered between different units of major

Portuguese hospital. It has been running for two

years, having integrated (searched, verified,

collected and stored) over 2 millions clinical

records.

In (Vieira-Marques, Robles et al. 2006) we

proposed a model for a medical-information-

gathering system that addresses issues such as inter-

institutional patient health data integration, the

retrieval of momentarily unavailable online remote

data, and secure data access and transport. In this

model, clinical events scheduling trigger agents to

gather information for a particular patient clinical

history, making that history as complete as possible

when those events occur. Integration efforts are

directed at clinical documents and not at the data

themselves.

Figure 1: MAID Mobile model description.

In this paper, we describe a pilot implementation of

the model which is focused in gathering information

from multiple institutions, aiming at creating a first

approach to a national wide Virtual Electronic

Patient Record (VEPR).

2 METHODS

The system design was divided in three phases. To

each phase corresponds one functional moment of

the system (Fig. 1).

2.1 Events Management Phase

This is the first phase and corresponds to the

management and scheduling of new events like

consultations or surgeries. These events trigger

actions for information search and retrieval. The

Scheduler Agent (SA) manages the events and a set

of information search and retrieval Mobile Agents

(MA). To each patient a set of remote places where it

is known to exist clinical information is stored. This

list will grow has new places with information are

discovered.

2.2 Information Search Phase

The second phase consists on a search performed by

MA that will move between remote systems looking

for information (exams, lab reports, etc) of a

designated patient. If any documents are discovered

references are collected and stored.

Besides clinical information, MA will ask for

additional platforms where it is known for the

patient data to exist. If other platforms are provided

then they are added to the agent itinerary.

2.3 Document Retrieval Phase

At this stage, the process of searching for

information references has ended and the collected

references have been stored locally for retrieval.

This retrieval will make use of basic remote transfer

services using ftp or http urls or trough interfacing

web services. After collection actions all the

information is made available to the end user.

3 RESULTS

A set of agents was implemented using JADE

platform and are described in the following sections

along with their interactions while pursuing the

designated objectives for each stage.

3.1 Scheduler Agent

This agent (SA) is in charge of scheduled events and

MA management.

When some event is scheduled the SA constructs

two itineraries, one with the remote systems that are

known to have patient information and other with

the complementary MAID systems retrieved from a

directory services agent. These itineraries are

delivered to two independent MA. Upon MA’s return

this agent is notified that the process is finished and

HEALTHINF 2008 - International Conference on Health Informatics

216

the MA have returned with the collected information.

This agent will also receive a message from the

Local Broker Agent (LBA) informing about

referenced documents retrieval success in order to

eliminate the event from its list.

3.2 Mobile Agent

The MA actions are crucial to the purpose of this

system as they are the ones in charge of discovering

and carrying new information about a given patient.

These agents’ actions are implemented using a

Finite State Machine (FSM) which describes its roles

and roles transitions during the different stages of

moving, remote systems interaction, etc.

As soon as the agent receives the message from

the SA, it saves the necessary data (itinerary and

patient id) and moves to the first place. While in the

remote platform, messages are exchanged with the

remote broker agent (RBA). This message exchange

has the purpose of obtaining the new local interface

role, allowing the MA to retrieve the document

references. During the initial exchange of messages

a simple authentication is performed.

Figure 2: Agent interactions taking place during

Information Search Phase.

As soon as the MA receives the role, it adds it to

its FSM defining the necessary additional state

transitions. From this moment on the agent is ready

to get existing data through the execution of the new

role. The retrieved data include a document

reference list and a list of platforms that the visited

system knows to have documents related to the

patient the MA represents. By doing so, it will

discover new sources of information. When the MA

finishes its itinerary it will go back to the original

platform. Upon arrival he notifies the SA of its return

and delivers the gathered information to the LBA.

3.3 Remote Broker Agent

This agent stays in the local system and acts as an

interface agent to incoming agents. It has to be

implemented in accordance to each platform

specificity, as all remote system at this point are

MAID systems they are implemented the same way

in all nodes. It is in charge of authenticating and

providing the interface behavior to the incoming

agent.

3.4 Local Broker Agent

This agent is in charge of receiving the result of the

references retrieval process. After receiving new

references from the MA, it stores them into the

database and starts the document retrieval process

trough the external services provided by each

platform. After the documents being retrieved they

are made available to the end user (Doctor, Nurse,

etc) through existing user interfaces. The documents

are saved in the file system digitally signed,

guaranteeing the documents integrity.

Figure 3: Agent interactions taking place during

Document Retrieval Phase.

3.5 Agent Interactions

The UML activity diagrams describe the interactions

between the different agents during information

references search (Fig. 2) and document retrieval

(Fig. 3).

4 DISCUSSION

This prototype extends the MAID systems by

implementing a set of interfacing agents enabling the

retrieval of remote documents. It addresses a

scenario where multiple MAID systems coexist. We

feel that multi-agent technologies can help

implement integration between heterogeneous

healthcare Information Systems in a satisfactory

manner.

4.1 Integration Achieved

Regarding ubiquity, our implementation enables the

exchange of information between each integrated

system in a similar way. All information is

A FIRST APPROACH FOR A REGIONAL WIDE VEPR

217

equidistant to each system. To face the problem of

latency of communications, or even momentary lack

of connectivity between different institutions when

the healthcare professional needs the patient

information, the system tries to collect useful patient

data a prior, i.e., before the user request. Scheduled

appointments trigger data discovery and report

collection. Agents facilitate ubiquity as they allow

MAID to operate in an asynchronous faction, which

is more powerful than other technology solutions

like web services that rely on synchronous

communication.

Regarding synchronization, our proposal takes

two different approaches. When integrating patient

data in an institution, MAID collects all updated

reports from the departmental IS so that its local

repository stays updated. In this scenario on the

other hand, when integration occurs between

different MAID systems our method is more like

lazy synchronization, i.e., the information is

updated only when a healthcare professional access

to the system is scheduled. Although, this approach

may raise problems in unscheduled accesses, we feel

that full synchronization stresses network resources

too much.

Although single system image is usually

regarded as being a user interface issue, MAID

delivers the same functionality of single image not

to human users but to requesting systems that may

have user interfaces themselves.

Consistency and integrity were not a major

concern in this implementation. Nevertheless it

should be noticed that MAID has already some data

quality checking in place, namely to detect wrong

patient identifications (Cruz-Correia, Vieira-

Marques et al. 2006). For full consistency checking,

it is essential to have documents introspection which

stresses the use of informatics standards like XML

to describe documents, or even healthcare related

standards like HL7-CDA or OpenEHR.

Unfortunately, the Portuguese reality is still far from

having healthcare IS that give access to their data

through this standards.

Transparency is very difficult to obtain when

dealing with documents generated in heterogeneous

IS, because they lack normalization regarding

presentation. To achieve a high level of transparency

it is essential to use structured documents (e.g.

XML).

4.2 Implementation Issues

By providing interfacing behaviours to incoming

mobile agent there is no need for complex

interactions reducing interface agent congestion

which would arise from multiple agents requesting

database queries.

REFERENCES

Berg, M. (2001). "Implementing information systems in

health care organizations: myths and challenges." Int J

Med Inf 64(2-3): 143-156.

Cruz-Correia, R., P. Vieira-Marques, et al. (2005).

"Integration of hospital data using agent technologies -

a case study." AI Communications 18: 191-200.

Cruz-Correia, R., P. Vieira-Marques, et al. (2006).

"Monitoring the integration of hospital information

systems: how it may ensure and improve the quality of

data." Stud Health Technol Inform 121: 176-82.

Lenz, R. and K. A. Kuhn (2002). "Integration of

Heterogeneous and Autonomous Systems in

Hospitals." Business Briefing: Data management &

Storage Technology.

Littlejohns, P., J. C. Wyatt, et al. (2003). "Evaluating

computerised health information systems: hard lessons

still to be learnt." BMJ 326: 860-3.

Moreno, A. (2006). "On the evolution of applying agent

technology to healthcare." IEEE Intelligent Systems

21(6): 8-10.

Vieira-Marques, P., S. Robles, et al. (2006). "Secure

Integration of Distributed Medical Data using Mobile

Agents." IEEE Intelligent Systems Journal 21(6): 47-

54.

HEALTHINF 2008 - International Conference on Health Informatics

218