Hybrid Simulation Approach for Prospective Assessment

of Mobile Stroke Units

∗

Anatoli Djanatliev

, Peter Kolominsky-Rabas

, Bernd M. Hofmann

, Axel Aisenbrey

and Reinhard German

Computer Networks and Communication Systems, Universit

at Erlangen-N

urnberg, Erlangen-N

urnberg, Germany

Interdisciplinary Centre for HTA and Public Health, Universit

at Erlangen-N

urnberg, Erlangen-N

urnberg, Germany

Siemens AG, Healthcare Sector, Clinical Competence Center Neuroscience, Erlangen, Germany

Siemens AG, Healthcare Sector, Siemens Healthcare Consulting, Erlangen, Germany

Keywords:

Healthcare, Simulation, Hybrid Simulation, System Dynamics, Agent-based Simulation, Prospective Health

Technology Assessment, Ischemic Stroke.

Abstract:

Technology innovations in health care offer high potentials for all stakeholders (e.g., patients, healthcare

providers and health industry), but the development phase of such innovations is often very expensive and

the effects are hardly predictable without a systematic strategy. The new interdisciplinary approach Prospec-

tive Health Technology Assessment (ProHTA) uses simulation techniques to indicate the effects of new health

technologies, early before the cost-intensive development process begins. Furthermore, ProHTA helps to de-

tect gaps and bottlenecks in the health system to catch potentials for new innovations. The scope of ProHTA

includes both a strategic aggregated level of analyses as well as an individual detailed level. This paper de-

scribes the use of hybrid simulation approaches, consisting of System Dynamics and Agent-Based Simulation,

to analyze the effects of an innovative stroke technology. We discuss an example of the prospective assess-

ment of Mobile Stroke Units within a Metropolitan Scenario. The project ProHTA is a part of the Centre of

Excellence for Medical Technology - Medical Valley EMN - and is supported by the German Federal Ministry

of Education and Research (BMBF), project grant No. 01EX1013B.

1 INTRODUCTION

The global market for medical technology products is

growing rapidly due to an increasing demand of in-

novative healthcare technologies. Presumably, some

reasons of this notice are the increasing life ex-

pectancy of the population, new technical opportu-

nities and an increasing complexity of the healthcare

delivery.

A signiﬁcant consequence of this trend is that the

healthcare industry has to handle even faster trade-

off decisions before developing new innovations in

medicine. To be more proﬁtable a new product must

have a short development phase, low costs and it

is also important that the expected revenue can be

reached. In addition, health technologies are safety-

critical, as the life quality can depend on the product’s

quality. Following this fact, proven evidence is nec-

∗

On behalf of the ProHTA Research Group.

essary and regulatory barriers have to be overcome.

Usually, such activities are time-consuming and their

results are difﬁcult to predict. Often there is also no

proven evidence available, so the development of new

innovations can degenerate into a risky, non-proﬁtable

project.

The main difference between the healthcare do-

main and other safety-critical areas is that many in-

volved stakeholders with different interests have to

be satisﬁed before a product can be placed on the

market. Some examples of them are patients, health-

care providers such as hospitals and outpatient depart-

ments, manufacturers in healthcare, technology de-

velopers, insurance companies, governments, regula-

tory institutions and academia.

According to the described situation, assessments

of health technologies are very complex and advanced

assessment methods are crucial to prevent disinvest-

ments and to make the product development more ef-

fective. All players must be considered for this pro-

357

Djanatliev A., Kolominsky-Rabas P., M. Hofmann B., Aisenbrey A. and German R..

Hybrid Simulation Approach for Prospective Assessment of Mobile Stroke Units.

DOI: 10.5220/0004029603570366

In Proceedings of the 2nd International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH-2012),

pages 357-366

ISBN: 978-989-8565-20-4

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

cess to reach an overall credibility of the product’s

quality and to detect all kinds of problems as early as

possible.

Three assessment methodologies for healthcare

technologies have been introduced until now. Health

Technology Assessment, Early Health Technology

Assessment and Horizon Scanning.

Health Technology Assessment (HTA) is an ap-

proach used after the market launch of a product when

evidence is available from studies. In accordance with

Goodman (Goodman, 2004), the main ﬁeld of HTA

is to inform, among others, regulatory agencies and

lawmakers about permission decisions for a commer-

cialization of a regarded innovation.

Early Health Technology Assessment (Early

HTA) tends to be an appropriate tool in cases where

the product is already in the development phase but

only limited evidence is available. Some examples for

such evidence levels are animal testing, early clinical

experience, or data from previous technology genera-

tions (Pietzsch and Pat

e-Cornell, 2008).

The impact of Horizon Scanning (HS) is the com-

parative assessment of similar technologies to observe

trends of possible disinvestments and impacts of new

healthcare technologies. Horizon Scanning Systems

focus on health technologies that are ready to enter the

market, i.e. in early post-marketing phases (Geiger-

Gritsch, 2008).

In many cases the technological progress is the

key factor for new products and there is only a little

consideration of possible future consequences. But

new innovations must reach positive assessment re-

sults before their impact can be seen by all stake-

holders as reasonable. Hence, an enormous number

of disinvestments can be expected, as established as-

sessment methods (HTA, Early HTA and HS) are just

used for products that already have passed the expen-

sive design and development process.

For that reason we call for advanced foresight

healthcare assessment methods which are applicable

early before high efforts and investments were made.

This is where our approach ProHTA can take an im-

portant role, because simulation techniques can be ap-

plied to assess new innovations prospectively.

2 RELATED WORK

There are already a couple of publications which are

presenting the use of simulation for the healthcare do-

main. Most of them are considering healthcare pro-

cesses and are modeled by the Discrete Event Sim-

ulation approach. Brailsford (Brailsford, 2008) dis-

cussed System Dynamics (SD) as tool for healthcare

modelers. The author presented several examples and

depicted some reasons for the growth in SD popular-

ity. Following this, some advantages of SD are lower

data requirements, the sight of the big-picture and in

particular very fast simulation runs.

In order to beneﬁt from high-level abstractions as

well as to allow more detailed modeling possibilities,

hybrid simulation techniques are gaining acceptance.

Heath et al. (Heath et al., 2011) discussed several

challenges and successes of cross-paradigm model-

ing. In particular, Discrete Event Simulation, System

Dynamics and Agent-Based Simulation approaches

were considered in detail. Though hybrid simulation

is not precisely deﬁned yet, there are however soft-

ware packages that allow multi-paradigm modeling.

3 PROSPECTIVE HEALTH

TECHNOLOGY ASSESSMENT

Prospective Health Technology Assessment (Pro-

HTA) is a new approach that extends the tool envi-

ronment of healthcare assessment methodologies and

ﬁlls the gap, mentioned previously. For this reason

the project is located within the Medical Valley EMN

(European Metropolitan Region Nuremberg) and is a

part of the Centre of Excellence for Medical Technol-

ogy.

The innovative intention of ProHTA is the assess-

ment of health technologies from many perspectives,

early before the development of a medical innovation

begins. It helps to detect potentials for process opti-

mization and allows learning about the inﬂuence of

a new technology on the established health system

structures. Innovation’s cost-effectiveness has to be

prospectively calculated as well as the impact on the

patient’s health.

Figure 1 summarizes the main prospective eval-

uation processes. In this respect, two questions are

central to ProHTA:

• What are the changes that result from the launch

of a new technology? - ProHTA will be able to

simulate and assess the effects of changes to pro-

cesses, which are introduced e.g. by a new tech-

nology. ProHTA will project the effects of inno-

vative health technologies on the quality and costs

of health care.

• What does a technology need to be like in order

to have a speciﬁc effect? - If certain input re-

quirements of a desired change are speciﬁed, the

ProHTA tools will enable outcome related con-

clusions concerning the required changes to the

process (hypothesis development). ProHTA will

SIMULTECH2012-2ndInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

358

envisage the effects of the potential efﬁciency en-

hancement on the health care system.

Figure 1: ProHTA overview.

Especially the second question allows ProHTA to

examine the existing health system and to ﬁnd bottle-

necks and weaknesses of currently applied practices.

This can lead to new ideas for health technologies

that are particularly based on desired effects and not

only on technical opportunities. In the course of the

method creation of ProHTA a new scientiﬁc service

platform has to be developed targeting on the scope

of the just described project.

The regarded challenges and questions are han-

dled together by an interdisciplinary team consisting

of experts from the areas of Public Health, Health

Technology Assessment, Clinical Medicine, Health

Economics and Outcomes Research, Medical Infor-

matics, Knowledge Management as well as Modeling

and Simulation. Furthermore, two representatives of

the healthcare industry also participate in the project.

In this paper we discuss the use of hybrid simula-

tion approaches, consisting of System Dynamics (SD)

and Agent-Based Simulation (ABS), for the new fore-

sight assessment method ProHTA. As the complexity

of the considered domain and the interdisciplinary co-

working is enormous, a structured methodology to-

wards a hybrid simulation model will be described

within the next section as well. The current focus

of the approach is the assessment of innovative tech-

nologies in management of acute stroke. We use Mo-

bile Stroke Units (MSU) within a Metropolitan area

as an innovative health technology to show exemplary

how hybrid simulation techniques can be applied to

the impact of ProHTA. Within the discussion section

we look ahead to the future work, including the vali-

dation of our approach and its application to oncology

diseases.

4 CONCEPTUAL MODELING

As already mentioned, many challenges have to be

mastered during the project realization phase. To

achieve an effective and intensive co-working of

all experts with different backgrounds, a structured

methodology is essential.

For that reason a dedicated Conceptual Modeling

Process (CMP) for ProHTA was deﬁned, classifying

the main steps towards a hybrid simulation model, be-

ginning at the domain experts’ knowledge collection.

It is largely based on current work of a research group

from the University of Warwick and Loughborough

University (Robinson, 2011; Kotiadis and Robinson,

2008).

As hybrid simulation builds the focus of this paper

and the CMP will be the topic of another publication,

we introduce in the following only some parts of it, as

it is important for the simulation.

The CMP distinguishes in particular between do-

main experts and technical experts and deﬁnes ﬁelds

of activity and intermediate interfaces for all of them.

Furthermore, the reality is summarized within the Do-

main World and the abstraction of it is contained in

the, so called, Model World. The signiﬁcant arti-

fact of the Domain World is the non-formal speciﬁ-

cation of the problem, represented by the Conceptual

Domain Model (CDM). This result is used by an it-

erative formalization process as input for the Model

World, according to create the Formal Conceptual

Model (FCM) which serves as a basis for the simu-

lation model and knowledge base.

Figure 2 depicts by solid lines the main steps of

the Conceptual Modeling Process towards a simula-

tion model. Dashed arrows represent the feedback

steps of iterative improvements.

5 USE-CASE: MOBILE STROKE

UNITS WITHIN A

METROPOLITAN REGION

Stroke is one of the leading causes of death and

disability and absorbs a considerable proportion of

healthcare budgets (Kjellstr

om et al., 2007). In Ger-

many an increase of 1.5 million stroke cases is esti-

mated in the next decade (Kolominsky-Rabas et al.,

2006). Due to an ageing population, increasing inci-

dence values could be expected in the future, as stroke

appears mainly at higher age groups. The costs for

treatment and rehabilitation in Germany are estimated

to be totally about 58 billion of euros in the period

2006-2015 (Kolominsky-Rabas et al., 2006).

HybridSimulationApproachforProspectiveAssessmentofMobileStrokeUnits

359

Figure 2: Conceptual modeling process of ProHTA.

One of the most frequent form of stroke is the is-

chemic stroke, caused by an oclusion of cerebral ar-

teries. Currently, an approved method for the treat-

ment is the intravenous thrombolysis by recombinant

tissue plasminogen activator (rtPA) (Fassbender et al.,

2003). However, even in specialized hospitals (Stroke

Units), only 7 to 10 percent of patients are treated by

this effective method (Heuschmann et al., 2010). The

main reason for such a bad rate is the applicability

of the thrombolysis only within 4.5 hours after the

obstruction’s occurrence and the time, elapsed during

the transport to the hospital (Purrucker and Veltkamp,

2011). A signiﬁcant increase of the rate from 3 per-

cent to 7 percent had been observed after the exten-

sion of the recommended time window from 3 hours

to 4.5 hours in 2008 (Purrucker and Veltkamp, 2011).

In case of an affection, nearly 1.9 million of neu-

rons can die per minute (Kuehn and Grunwald, 2011).

Hence, the “time is brain” concept has to be applied

to the extremely time-critical treatment of stroke. Fol-

lowing this fact, particularly innovations in stroke

treatment are necessary which are able to reduce the

call-to-therapy-decision time.

Two German research groups from Saarland and

Berlin (Walter et al., 2010; Ebinger et al., 2012) are

working with their partners on methods targeting at

the transfer of the thrombolytic inpatient treatment to

the pretreatment phase. In that case the therapy can be

applied before the time-intensive transfer to the hos-

pital and an increase of the number of patients, treated

by thrombolysis can be expected. The main problem

of this idea is the important exclusion of an intracere-

bral haemorrhage, before applying rtPA. This can be

done by laboratory analyses and Computer Tomogra-

phy (CT), usually installed in hospitals. To prevent a

loss of crucial time, both research groups developed

similar prototypes, the Mobile Stroke Unit (MSU) and

the Stroke-Einsatz-Mobil (STEMO). Such a vehicle

extends the standard emergency equipment by a CT

and further tools for a rapid diagnostic decision on-

site at stroke occurrence location.

In an early phase, ﬁrst trials have shown that

a shortened call-to-therapy-decision-time of approx-

imately 35 minutes is not a vision (Kuehn and Grun-

wald, 2011). To inspect other relevant effects (e.g.

long-term cost-effectiveness, application in other re-

gions) more trials are necessary, so that further time

and cost investments have to be made. This is where

ProHTA can provide an early analysis by hybrid sim-

ulation.

6 HYBRID SIMULATION

For the purpose of the Prospective Health Technol-

ogy Assessment simulation has been identiﬁed to be

an appropriate tool. Customized large scale simula-

tion models are crucial to handle complex questions

in situations where an innovative technology possibly

hasn’t been developed yet. Depending on the point

of view, ProHTA must be able to answer specialized

questions (e.g. effectiveness of a new innovation)

SIMULTECH2012-2ndInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

360

as well as to calculate global and long-term conse-

quences. Hence, large scale models are necessary to

fulﬁll these requirements. Furthermore, separate data

management and data quality components are crucial

to handle a couple of different input data formats in

an efﬁcient way (Baumg

artel and Lenz, 2012).

We need methods to create models on an abstract,

macroscopic level (e.g. economic ﬂows, population

dynamics) where data about details is not available,

or even is not necessary. In our approach System Dy-

namics (SD) (Forrester, 1999) was qualiﬁed as an ap-

propriate tool for this kind of simulation. To model

individual aspects of patient’s workﬂows and its be-

havior, the Agent-Based Simulation (ABS) approach

has been selected. New technological innovations can

also be modeled by the ABS approach to allow be-

havioral changes, according to inspect their impacts

on output values.

Each assessment scenario is individual. For this

reason we propose to combine the SD and the ABS

approach into a common hybrid simulation using an

adequate ratio according to the desired simulation

level and high simulation performance. This trade-

off decision can be handled by following a top-down

approach. In that case the problem environment can

be modeled as deeply as needed by abstract methods

and a special simulation part will be framed-out and

realized by the ABS approach.

The idea of hybrid simulation is getting more

and more popular in current research (Heath et al.,

2011; Brailsford et al., 2010). There are many soft-

ware packages that allow to create simulations by SD,

Discrete Event Simulation (DES) or ABS models,

but most of them support only one of the presented

approaches. AnyLogic (XJ Technologies Company

Ltd., 2012) offers the power to combine different sim-

ulation paradigms within a common simulation envi-

ronment. This is why this tool is qualiﬁed to create

models within the scope of ProHTA.

6.1 Modular Environment

In accordance to the Conceptual Modeling Process

(Section 4), the CDM represents a non-formal de-

scription of the collected experts’ knowledge. The

FCM is central to develop simulation models within

the context of ProHTA.

We have to deal with different, interacting simula-

tion parts, such as money ﬂows, healthcare structures,

disease, and population development. To make mod-

els reusable for further assessment scenarios and to

master the complexity, modularization is important.

Figure 3 depicts the identiﬁed modules within the

scope of ProHTA.

Figure 3: Overview of the modules.

Population Dynamics summarizes models and pa-

rameters that handle with the development of demo-

graphic structures, e.g. birth rate, mortality, immigra-

tion and emigration.

Disease Dynamics includes generic model parts

that deal with illness parameters, e.g. incidence,

prevalence, case fatality rate and remission.

Economics combines the dynamics of money

ﬂows of the health system. Statutory and private

health insurance can be modeled there as well as the

long-term care system.

Health Care is a module where especially work-

ﬂows of prevention, pre-treatment, treatment and

post-treatment of a regarded disease can be modeled.

Chahal and Eldabi (Chahal and Eldabi, 2008) pro-

posed a process environmental format for hybrid sim-

ulations. In our hybrid simulation approach for Pro-

HTA, SD models can be used to build the environment

of a simulation. An important simulation part can be

modeled by ABS approaches and used as core within

the process environmental format.

6.2 Scenario Description

Before we set the focus on our model, the scenario

with Mobile Stroke Units within a Metropolitan re-

gion will be shortly presented in the following:

• Stroke scenario with usage of a predeﬁned num-

ber of 10 MSUs.

• A German Metropolitan region, represented by

Berlin, with approximately 3.416.000 people is

modeled using data from the statistical calcula-

tions of (Amt f

ur Statistik Berlin-Brandenburg,

2010).

• The regional distribution of people is done by in-

formation about the district density.

• MSUs are randomly distributed within the region

boundaries.

• People can get stroke and call the Rescue Service

(RS).

• RS decides whether an MSU can be sent (e.g.

HybridSimulationApproachforProspectiveAssessmentofMobileStrokeUnits

361

Figure 4: Mobile stroke units within a metropolitan scenario.

there are free MSUs available) to the affected pa-

tient.

• During the affection phase, people pass through

diagnostic and therapy workﬂows and lose life

quality according to a therapeutic effectiveness.

• In case of using an MSU, thrombolytic therapy

can be started onsite at patient’s location, after

a mobile CT and laboratory diagnose were per-

formed.

• Calculated costs are drawn from ﬁnancial bud-

gets.

• Demographical information is used to reproduce

the population development.

• Prevalence is used to separate the population in

affected and not-affected parts.

• Incidence is used as a dynamic affection rate.

For assessment purposes the same scenario with-

out MSUs will be used afterwards. It allows to make

the effects of a new intervention (MSUs) visible.

6.3 Simulation Model

6.3.1 Overview

The simulation is started within an initialization phase

and proceeds with a forecasting phase after a couple

of data has been produced.

There are two types of agents modeled within the

hybrid simulation of our MSU scenario, the individ-

ual behavior of persons and Mobile Stroke Units. De-

mographic development is reproduced by a separate

System Dynamics model (Population Dynamics) as

well as money ﬂows of the health system.

Figure 4 depicts a screenshot of a running simu-

lation. Small points represent person agents (number

1). Changing the color from blue to red a patient be-

comes affected. Yellow bullets with black borders are

a representation of MSU agents (number 2). During

simulation runs they are colored in red, if an MSU is

busy to see an overall utilization of the new technol-

ogy. MSUs pick up patients and move them to the

next hospital.

6.3.2 Agents

Age, risk factor (RF), gender and life quality (LQ)

are important attributes of a person and are sampled

initially. A normal risk is represented by RF = 1, high

risk by RF > 1 and low risk is a value RF < 1 ∧ RF >

0. LQ is represented by a value between 0 and 100.

Dependent on the therapy success, the patient loses

more or less LQ points.

Time until the next occurrence of stroke symp-

toms is calculated by an appropriate function. An

annualEvent is used to increase patient’s current age

and to run a death calculation function. The behavior

SIMULTECH2012-2ndInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

362

is represented by the left hand state chart of Figure 5

with three states (normal, symptoms, affected).

Figure 5: Behavior state charts of agent person (left) and

MSU (right).

The state chart of the agent type MSU is depicted

on the right hand of the Figure 5 and includes two

states (free, busy). The variable timeApprToScene is

used to simulate the delay of an MSU until arrival at

patient’s location. In a further step it can be used to

reproduce trafﬁc by an additional simulation part.

The health care structure is modeled within a sep-

arate state chart. At the beginning, each person is lo-

cated in the state noHealthCare. After the occurrence

of suitable events, a patient changes to an appropri-

ate composite state, such as prevention, pre-treatment,

treatment and post-treatment that are partially pre-

sented in the following.

Each step that produces costs (e.g. special treat-

ments, materials, staff employment) collects them in

a global variable within the economics System Dy-

namics model. The calculated number is pulled off

from the corresponding budgets once-a-year.

6.3.3 Prevention

As prevention interventions for stroke are not usual, a

simple part of the model takes into account only weak

preventive inﬂuences, e.g. healthy lifestyle. Preven-

tive check-ups are simulated in undetermined cycles

with a minimum of ﬁve years elapsed since the last

prevention.

The time until next stroke symptoms occurrence is

calculated by a multiplication of the remaining time

and the individual risk factor (RF) of a person. The

higher a RF, the higher the prevention effect and the

earlier a prevention, the higher is the gain of time.

6.3.4 Pre-treatment

After the expiration of the time until next symptoms,

an event is ﬁred and the concerned person enters the

symptoms state and starts the workﬂow of the pre-

treatment phase within the HealthCare state chart, ﬁg-

ured in 6.

Figure 6: Agent-based model of the pre-treatment phase.

According to the workﬂow an emergency call is

modeled ﬁrst. After talking to the dispatcher a ﬁrst

diagnosis is made and a rescue service is alarmed to

move to the patient. Due to the assessment of MSUs,

the dispatcher searches for free Mobile Stroke Units.

The more of them are initially available, the higher is

the probability to get one free. Otherwise, a normal

emergency doctor will be sent.

In case of MSU usage, a CT exam can be per-

formed at the patient’s location in the pre-treatment

phase, as described in section 5.

Following our assessment scenario, the branches

where MSU are involved are set to 0 during the com-

parative simulation runs.

6.3.5 Treatment

Due to the arrival in the hospital the treatment phase

starts by an invocation of the composite state treat-

ment. Diagnosis and therapy steps which were al-

ready performed during the pre-treatment phase are

HybridSimulationApproachforProspectiveAssessmentofMobileStrokeUnits

363

Figure 7: Mean estimated life quality without/with 10 MSUs.

skipped and a best-case workﬂow, modeled by our do-

main experts, is traversed by the patient.

6.4 Simulation Results

In our exemplary use-case of Mobile Stroke Units we

noticed some trends due to the implementation of 10

MSUs in a metropolitan area. According to the used

input parameters that were partially estimated by do-

main experts, we monitored some of the expected re-

sults.

Figure 7 shows two plots of the estimated mean

life quality. On the left hand we can see graphs that

were calculated by a simulation run without MSU

usage; on the right hand a scenario with 10 Mobile

Stroke Units was applied. According to the results we

can notice that persons with stroke affection achieve

a higher mean life quality after the launch of the re-

garded innovation. The main reason for this is the

possibility to apply the thrombolytic therapy imme-

diately at the stroke occurrence location and an in-

tervention can be done within the important time-

window of 4.5 hours.

In our model we also aggregated costs caused by

MSU usage to show exemplary economic assessment

calculations. To check if a new technology is reason-

able, a Cost-Effectiveness-Analysis (CEA) was com-

puted by the equation 1 during simulation runs.

CEA =

MSU

− K

NoMSU

∆LQ

NoMSU

− ∆LQ

MSU

(1)

We regarded life quality as effectiveness parame-

ter to determine the effect of MSUs. Within the equa-

tion the following variables are deﬁned as follows:

MSU

− K

NoMSU

(cost difference between MSU usage

and conventional therapy workﬂows), ∆LQ

NoMSU

and

∆LQ

MSU

are mean life quality losses of stroke patients

without/with MSU implementation.

Graph 8 shows an example result of a Cost-

Effectiveness-Analysis. The plot suggests that an in-

crease of the mean quality of life of the population of

one point comes with a cost of approximately 12,000.

It also shows the effects of the initialization phase

which have been alleviated in the forecasting phase

starting around the year 2010.

7 DISCUSSION

ProHTA is a new approach that uses simulation tech-

niques to assess health technologies early before the

cost-intensive development process starts. It allows to

evaluate healthcare innovations within the context of

the health system and to ﬁnd gaps and bottlenecks that

can lead to new ideas for innovations. We identiﬁed

hybrid simulation, consisting of the System Dynam-

ics and the Agent-Based Simulation, to be an appro-

priate modeling approach for large scale simulations.

This method enables to solve problems on a macro-

scopic level, e.g. disease dynamics in a global con-

text, as well as on a detailed, behavioral microscopic

level. A further beneﬁt of this approach within the

scope of ProHTA is the possibility to build models

top-down, beginning from high abstractions and go-

ing more in detail to frame-out individual workﬂows.

To make models capable for reuse and to master com-

Figure 8: Example of a cost-effectiveness-analysis (CEA).

[y-axis: costs per one life quality point, x-axis: time].

SIMULTECH2012-2ndInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

364

plexity, we designed a modular and generic environ-

ment that is also eligible for ﬂexible model changes.

As an effective co-working of interdisciplinary ex-

perts is crucial to get useful results, a conceptual mod-

eling process was developed. The main beneﬁts are in

particular well-deﬁned areas of activity for all experts

and the capability to proceed in a structured way.

To evaluate our methods, an exemplary use-

case scenario of an innovative stroke treatment ap-

proach, represented by Mobile Stroke Units within

a Metropolitan Scenario, had been implemented, us-

ing the simulation software AnyLogic (XJ Technolo-

gies Company Ltd., 2012). This tool is predestinated

for multi-method simulation paradigms. The project

procedure strictly followed the CMP and an overall

expert-credibility has been achieved.

There are still many challenges to master in the fu-

ture. Real data from the Stroke Register of Erlangen

(ESPro) will be used to validate the model using other

stroke use-cases and to asses interventions whose ef-

fects are already attestable by evidence data. A fur-

ther complex task will be the application of our hy-

brid simulation approach to other diseases, especially

within the domain of personalized medicine. As the

ProHTA research group includes oncology experts,

cancer diseases will be the focus of further work.

Some technical challenges are also still remaining.

Some of them are simulation performance, scalabil-

ity of models, requesting data (semi-)automated from

the data management component and further hybrid

simulation research for combination of SD and ABS

models.

ACKNOWLEDGEMENTS

Prospective Health Technology Assessment (Pro-

HTA) is funded by the German Federal Ministry of

Education and Research (BMBF) as part of the Na-

tional Cluster of Excellence Medical Technology -

Medical Valley EMN (Project grant No. 01EX1013).

REFERENCES

Amt f

ur Statistik Berlin-Brandenburg (2010). Die kleine

Berlin–Statistik 2010.

Baumg

artel, P. and Lenz, R. (2012). Towards data and

data quality management for large scale healthcare

simulations. In Conchon, E., Correia, C., Fred, A.,

and Gamboa, H., editors, Proceedings of the Interna-

tional Conference on Health Informatics, pages 275–

280. SciTePress - Science and Technology Publica-

tions. ISBN: 978-989-8425-88-1.

Brailsford, S. C. (2008). System Dynamics: What’s in it for

Healthcare Simulation Modelers. In Proceedings of

the 2008 Winter Simulation Conference, pages 1478–

1483. Winter Simulation Conference 2008.

Brailsford, S. C., Desai, S. M., and Viana, J. (2010). To-

wards the holy grail: Combining system dynamics and

discrete-event simulation in healthcare. In Proceed-

ings of the 2010 Winter Simulation Conference, pages

2293–2303. Winter Simulation Conference 2010.

Chahal, K. and Eldabi, T. (2008). Applicability of hybrid

simulation to different modes of governance in UK

healthcare. In Proceedings of the 2008 Winter Sim-

ulation Conference, pages 1469–1477. Winter Simu-

lation Conference 2008.

Ebinger, M., Rozanski, M., Waldschmidt, C., Weber, J.,

Wendt, M., Winter, B., Kellner, P., Baumann, A.,

Malzahn, U., Heuschmann, P., Fiebach, J., Endres,

M., and Audebert, H. (2012). PHANTOM-S: the pre-

hospital acute neurological therapy and optimization

of medical care in stroke patients – study. Interna-

tional Journal of Stroke. Feb 2.

Fassbender, K., Walter, S., Liu, Y., Muehlhauser, F.,

Ragoschke, A., Kuehl, S., and Mielke, O. (2003).

”Mobile Stroke Unit” for Hyperacute Stroke Treat-

ment. Stroke, 34(6):44e–44.

Forrester, J. W. (1999). Industrial Dynamics. Pegasus Com-

munications, Waltham and MA.

Geiger-Gritsch, S. (2008). Horizon Scanning in Oncology:

Concept Development for the Preparation of a Hori-

zon Scanning System in Austria. Ludwig Boltzmann

Institute for Health Technology Assessment.

Goodman, C. S. (2004). HTA 101: Introduction to Health

Technology Assessment. National Information Center

on Health Services Research and Health Care Tech-

nology (NICHSR). The Lewin Group.

Heath, S. K., Brailsford, S. C., Buss, A., and Macal, C. M.

(2011). Cross-paradigm simulation modeling: chal-

lenges and successes. In Proceedings of the 2011 Win-

ter Simulation Conference, pages 2788–2802. Winter

Simulation Conference 2011.

Heuschmann, P., Busse, O., Wagner, M., Endres, M., Vill-

ringer, A., R

other, J., Kolominsky-Rabas, P., and

Berger, K. (2010). Frequency and Care of Stroke in

Germany. Aktuelle Neurologie, 37(07):333–340.

Kjellstr

om, T., Norrving, B., and Shatchkute, A. (2007).

Helsingborg Declaration 2006 on European Stroke

Strategies. Cerebrovascular Diseases, 23(2-3):229–

241.

Kolominsky-Rabas, P. L., Heuschmann, P., Marschall, D.,

Emmert, M., Baltzer, N., Neundoerfer, B., Sch

offski,

O., and Krobot K.J. (2006). Lifetime Cost of Ischemic

Stroke in Germany: Results and National Projections

From a Population-Based Stroke Registry: The Erlan-

gen Stroke Project. Stroke, 37(5):1179–1183.

Kotiadis, K. and Robinson, S. (2008). Conceptual mod-

elling: Knowledge acquisition and model abstraction.

In Proceedings of the 2008 Winter Simulation Confer-

ence, pages 951–958. Winter Simulation Conference

2008.

HybridSimulationApproachforProspectiveAssessmentofMobileStrokeUnits

365

Kuehn, A. and Grunwald, I. (2011). Das Konzept der Mo-

bilen Stroke Unit (MSU). Der Radiologe, 51(4):261–

262.

Pietzsch, J. B. and Pat

e-Cornell, M. E. (2008). Early tech-

nology assessment of new medical devices. Inter-

national Journal of Technology Assessment in Health

Care, 24(01):36–44.

Purrucker, J. and Veltkamp, R. (2011). Thrombolysis Be-

yond the 3-h Time Window. Aktuelle Neurologie,

38(09):494–504.

Robinson, S. (2011). Choosing the right model: Con-

ceptual modeling for simulation. In Proceedings of

the 2011 Winter Simulation Conference, pages 1428–

1440. Winter Simulation Conference 2011.

Walter, S., Kostpopoulos, P., Haass, A., Helwig, S., Keller,

I., Licina, T., Schlechtriemen, T., Roth, C., Pa-

panagiotou, P., Zimmer, A., Vierra, J., K

orner, H.,

Schmidt, K., Romann, M.-S., Alexandrou, M., Yil-

maz, U., Grunwald, I., Kubulus, D., Lesmeister, M.,

Ziegeler, S., Pattar, A., Golinski, M., Liu, Y., Volk,

T., Bertsch, T., Reith, W., Fassbender, K., and Noor,

A. M. (2010). Bringing the Hospital to the Patient:

First Treatment of Stroke Patients at the Emergency

Site. PLoS ONE, 5(10):e13758.

XJ Technologies Company Ltd. (2012). AnyLogic -

www.xjtek.com.

SIMULTECH2012-2ndInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

366