The Role of Modeling and Simulation in Coordination of Health

Care

Bernard P. Zeigler

Arizona Center for Integrative Modeling and Simulation, U. Arizona, Tucson, AZ, U.S.A.

RTSync Corp Phoenix, AZ, U.S.A.

Keywords: System of Systems, Coordination of Activities, Healthcare Coordination, DEVS Modeling and Simulation.

Abstract: US healthcare, the most expensive in the world, has been diagnosed as an assemblage of uncoordinated

component systems embedded in a market economy that promotes independent pricing with few points of

global control over delivered quality of care and cost. Stimulated by the Affordable Care Act and other

initiatives, efforts are underway to increase the level of information technology (IT) to improve patient

record keeping and portability as well as to price services based on performance rather than amount

provided. Yet such an IT infrastructure by itself will not provide significantly greater component

coordination since it does not provide transparency into the threads of transactions that represent patient

treatments, their outcomes, and total costs. In the traditional formulation of the coordination problem, the

goal of the system-of-systems conflicts with those of its components. In contrast, our concern here is the

coordination of activities among disconnected provider systems to deliver the appropriate services to an

individual client. In this paper, we discuss the Pathways Coordination model, a generic construct that

enforces threaded distributed tracking of individual patients experiencing certain pathways of intervention,

thereby supporting coordination of care and fee-for-performance based on end-to-end outcomes. DEVS-

based Modeling and Simulation methodology is discussed as the means to design, simulate, and implement

such client-based coordination in systems engineering.

1 HEALTH CARE REFORM – A

SYSTEM OF SYSTEMS

COORDINATION PROBLEM

An AHRQ/NSF workshop (AHRQ/NSF,2009)

envisioned an ideal health care system that is unlike

today’s fragmented, loosely coupled, and

uncoordinated assemblage of component systems.

The workshop concluded that, “An ideal (optimal)

health care delivery system will require methods to

model large scale distributed complex systems.”

Improving the health care sector presents a

challenge in that the optimization cannot be

achieved by sub-optimizing the component systems,

but must be directed at the entire system itself.

Reforming such a system requires methods to

model large scale distributed complex systems

using net-centric systems of systems engineering

approaches (Jamshidi 2008, Mittal et al., 2012,

PCAST 2014). Porter and Teisberg (2006) advocate

radical reform of health care that requires that

physicians re-organize themselves into Integrated

Practice Units (IPUs) moving away from care that is

currently based on specialties with associated

hospital departments. An IPU is centred on a

medical condition defined as an interrelated set of

patient medical circumstances best addressed in an

integrated way. Distinct from such clinical care is

extra-clinical care – the care needed outside the

hospital. People with multiple health and social

needs are high consumers of health care services,

and are thus drivers of high health care costs. The

ability to provide the right information to the right

people in real time requires a system-level model

that identifies the various community partners

involved and rigorously lays out how their

interactions might be effectively coordinated to

improve care for the neediest patients that cost the

most.

A useful abstraction comprehends the healthcare

system as an interaction of individual patients, a

variety of care providers, a set of payers, and a

billing system that records patient-provider

transactions and enables payer-provider fee-for-

service transactions. Stimulated by the Affordable

Care Act and other initiatives, efforts are underway

P.Zeigler B.

The Role of Modeling and Simulation in Coordination of Health Care.

DOI: 10.5220/0006813800010001

In Proceedings of the 4th Inter national Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2014), pages 5-16

ISBN: 978-989-758-038-3

to increase the level of information technology (IT)

to improve patient record keeping and portability as

well as to price services based on quality of service

rather than amount of service provided. Yet such an

IT infrastructure by itself will not provide

significantly greater coordination since it does not

provide transparency into, and management of the

threads of transactions that represent patient

treatments, their outcomes, and total costs.

Healthcare has been compared to manufacturing

with the idea that many of the same techniques can

be transferred to it. However, complex patient

flows, numerous human resources, dynamic

evolution of patient’s health state motivated

Augusto and Xie (2014) to develop Petri-net-based

software for modeling, simulation, and activity

planning and scheduling of health care services.

Their goal was to provide a mathematical

framework to design models of a wide range of

medical units of a hospital in order to model and

simulate a wide range of healthcare services and

organizations and to support such design with a

Unified Modeling Language (UML) / business

process modeling (BPM) interface for decision-

makers. In contrast, our concern here is not within

the hospital but at the System-of-Systems (SoS)

level where hospitals interact with other

components such as physicians, community

workers, social services and health plan payers.

At the SoS level, care coordination/ is the

organization of care activities among the individual

patient and providers involved in the patient’s care

to facilitate the appropriate delivery of health care

services. Craig et al. (2011) present a care

coordination framework aimed at improving care at

lower cost for people with multiple health and

social needs. Although such a framework provides a

starting point, it does not afford a rigorous

predictive model that takes account of emerging

health information networks and electronic medical

records. The Pathways Community HUB Model is a

delivery system for care coordination services

provided in a community setting (AHRQ, 2011).

The model is designed to identify the most at-risk

individuals in a community, connect them to

evidence-based interventions, and measure the

results (Zeigler et al. 2014). Community care

coordination works at the SoS level to coordinate

care of individuals in the community to help address

health disparities including the social barriers to

health. The Pathways Community HUB model is a

construct that enforces threaded distributed tracking

of individual clients experiencing certain pathways

of intervention, thereby supporting coordination of

care and fee-for-performance based on end-to-end

outcomes. As an essential by-product, the Pathway

concept also opens up possibilities for system level

metrics that enable more coherent transparency of

behavior than previously possible, therefore greater

process control and improvement re-engineering,

The objective of this paper is to present a

concept of Coordination Model that abstracts

essential features of the Pathways Community HUB

Model so that the kind of coordination it offers can

be appreciated, and employed, in a general SoS

context. This will allow us to develop a Modeling

and Simulation (M&S) framework to design, test,

and implement such coordination models in a

variety of SoS settings, exemplified by healthcare,

that present the issues that such coordination

models address. We discuss the associated concepts

and show how the Discrete Event System

Specification (DEVS) formalism is an appropriate

vehicle for their representation and implementation

in the MS4 Modeling and Simulation Environment

(Zeigler and Sarjoughian, 2013). We close with

comments on how the environment and tools can

play a role in the evolution of the Pathways

Coordination model as it is applied to healthcare

and other settings.

2 SYSTEM THEORY AND

SYSTEM-OF-SYSTEMS (SOS)

We begin with a brief review of systems theory,

systems-of-systems, and DEVS as they are relevant

to the type of coordination discussed here.

Application of the System of Systems Engineering

(SoSE) concept to healthcare recognizes that it

includes myriads of stakeholders involving multiple

large scale concurrent and complex systems that are

themselves comprised of complex systems

(Jamshidi, 2008, Wickragemansighe, et. al, 2008.)

Systems theory, especially as formulated by

Wymore (Ören and Zeigler, 2012), provides a

conceptual basis for formulating the coordination

problem of interest here. Systems are defined

mathematically and viewed as components to be

coupled together to form a higher level system, the

SoS.

2.1 Wymore’s Mathematical System

Framework

As illustrated in Figure 1, Wymore’s (1967)

systems theory mathematically characterizes:

Figure 1: Wymore's System Theory

Systems as well defined mathematical objects

characterizing “black boxes” with structure and

behavior.

Composition of Systems – constituent systems

and coupling specification result in a system, called

the resultant, with structure and behavior emerging

from their interaction.

Closure under coupling – the resultant is a

well-defined system just like the original

components.

2.2 System-of-Systems

As illustrated in Figure 2, a System of Systems

(SoS) is a composition of systems, where often

component systems have legacy properties e.g.,

autonomy, belonging, diversity, and emergence

(Boardman and Sauser, 2006). In this view, an SoS

is a system with the distinction that its parts and

relationships are gathered together under the forces

of legacy (components bring their pre-existing

constraints as extant viable systems) and emergence

(it is not totally predictable what properties and

behavior will emerge.) Here in Wymore’s terms,

coupling captures certain properties of relevance to

coordination, e.g., connectivity, information flow,

etc. Structural and behavioral properties provide

the means to characterize the resulting SoS, such as

fragmented, competitive, collaborative, coordinated,

etc.

Figure 2: System of Systems.

In the traditional formulation of the coordination

problem, each system has a goal and often the goal

of the SoS conflicts in part with those of the

components. Coordination is then conceived as a

mechanism to achieve optimal alignment of

component goals to the overall goal (Mesarovic,

1970). In contrast, as mentioned above, our concern

here is the organization of activities among

individual clients and service providers to

coordinate the appropriate delivery of services.

Although salient in healthcare, this concept of

coordination is applicable to many situations where

multiple providers offer multiple services to

multiple clients.

2.3 Discrete Event Systems

Specification (DEVS) Modeling

and Simulation Framework

The DEVS formalism (Zeigler,.Kim and Praehofer,

2000), based on systems theory, provides a

framework and a set of modeling and simulation

tools to support Systems concepts in application to

SoSE (Mittal et al, 2008). A DEVS model is a

system-theoretic concept specifying inputs, states,

outputs, similar to a state machine. Critically

different however, is that it includes a time-advance

function that enables it to represent discrete event

systems, as well as hybrids with continuous

components, in a straightforward platform-neutral

manner. DEVS provides a robust formalism for

designing systems using event-driven, state-based

models in which timing information is explicitly

and precisely defined. Hierarchy within DEVS is

supported through the specification of atomic and

coupled models. Atomic models specify behavior of

individual components. Coupled models specify the

instances and connections between atomic models

and consist of ports, atomic model instances, and

port connections. The input and output ports define

a model’s external interface, through which models

(atomic or coupled) can be connected to other

models.

Figure 3: DEVS Formulation of Systems-of-Systems.

As illustrated in Figure 3, based on Wymore’s

systems theory, the DEVS formalism

mathematically characterizes:

DEVS Atomic and Coupled Models specify

Wymore Systems.

Composition of DEVS Models – component

DEVS and coupling result in a Wymore system,

called the resultant, with structure and behavior

emerging from their interaction.

Closure under coupling – the resultant is a

well-defined DEVS just like the original

components.

Hierarchical Composition – closure of

coupling enables the resultant coupled models to

become components in larger compositions.

2.4 Client-Oriented Coordination of

Cross-System Transactions

In the kind of coordination considered here, there

are multiple service providers (component systems)

whose activities must be brought together in

different ways to serve different clients. In the as-is

situation, a client is to a large extent responsible for

selecting, sequencing, and scheduling encounters

with providers. Since multiple activities are located

in different component systems, the client needs to

traverse several activities across different systems

to complete a cross-system transaction. Thus an

adequate coordination model is characterized by the

following requirements:

 Coordination design must define cross-

system transitions and criteria for their

successful completion,

 One or more cross-system transactions

may be assigned to a client

 A coordination agent must aim to assure

that clients will successfully complete their

assigned transactions

 Coordination tracks the completion state

and provides accountability for

success/failure of the client and

coordination agent in completing assigned

transactions

 Coordination tracks the cost of cross-

system transaction by accumulating the

costs of activities involved in it.

2.5 Pathways as Coordination Models

Viewed as coordination models as just defined,

Coordination Pathways provide concrete means to:

 Define steps in terms of goals and subgoals

along paths to complete cross-system

transactions

 Test for achievement and confirmation of

pathway goals and subgoals

Figure 4: Methodology for Coordination Systems

Engineering.

 Track, and measure progress of, clients

along the pathways they are following

 Maintain accountability of the compliance/

adherence of the individual and responsible

coordination agent

An information technology implementation of

such Pathways can provide abilities to:

 Query for the state of a client on a pathway

 Query for population statistics based on

aggregation of pathway states for

individuals

 Support Time-Driven Activity-based

costing (Kaplan and Anderson, 2004)

based on pathway steps and their

completion times.

3 M&S METHODOLOGY FOR

COORDINATION MODELING

A methodology for coordination systems

engineering follows along the lines of net-centric

system of systems engineering with DEVS-based

modeling and simulation methodology (Zeigler and

Sarjoughian, 2013, Mittal et al. 2012.) As illustrated

in Figure 4, an SoS can be abstracted to a

simulation model which can be used to test the

developed coordination models. Such models can

be derived by abstracting the features (activities,

services, etc.) of the component systems that are

relevant to defining coordination pathways for

cross-system transactions of interest. Then after

virtual testing in the SoS simulation, the same

pathway models can be implemented in net-centric

information technology using the model-continuity

properties of the DEVS framework (which allows

simulation models to be executed in real-time as

software by replacing the underlying simulator

engine.) The following sections are in tune with this

approach.

4 DEVS FORMALIZATION OF

COORDINATION PATHWAYS

Formalization provides a firm basis for capitalizing

on the transparency that is afforded by the Pathways

Community HUB Model which enforces threaded

distributed tracking of individual clients

experiencing Pathways of intervention. Zeigler et al.

(2014) represented such pathways as DEVS Atomic

Models with implementation in the form of an

active calendar that combines event-based control

(Zeigler, 1989), time management, and database

capabilities. Here we will specify more precisely the

coordination pathway models just defined as a sub-

class of DEVS models. Further, such DEVS

Pathways can become components of coupled

models thereby enabling activation of successors

and sharing of information. Zeigler et al. (2014)

represented steps in a Pathway as states in its

associated atomic model. Such a representation can

constrain steps to follow each other in proper

succession with limited branching as required;

external input can represent the effect of a transition

from one step to next due to data entry. Moreover,

temporal aspects of the Pathways, including

allowable duration of steps can be directly

represented by the DEVS atomic model’s

assignment of residence times in states.

4.1 Atomic Pathways Models

Three aspects of Atomic Pathway models to note

are:

 Their primary role is to request and receive

data about a main goal and benchmarks (or

subgoals) accomplishment – we will call

these Questions and Answers

 Bounded times are given for answers to be

received

 Accomplishment of the main goal is

decidable after a finite time in the sense

that the model is guaranteed to wind up

(and remains) in one of three classes of

states: known success, known failure, or

incomplete. In the last type, the model

explicitly states that it is unknown whether

the goal has been achieved or not.

An example of an Atomic Model representing a

Pathway with one goal (Figure 5) starts in state WA

Figure 5: A Single Question Pathway Atomic Model.

(for waitForActivate) which is passive (its time

advance, ta is infinity). When an Activate is

received (input ports are noted by ?,output ports by

!), the model transitions to the Initialization state, I

which is a transient state (ta = 0.) This state

immediately outputs the question, GoalReached and

transitions to the state WG (waitForGoal.) In this

state, the model can receive answers Yes or No and

eventually enter passive states S (Success) and F

(Failed) resp. (S is entered after an Activate output

is generated from state SY.) However, WG has a

finite time advance, so that it transitions to states

Inc (incomplete) if it does not receive one of the

Yes or No answers within this interval. Since Inc is

a passive state, it is easy to see that, as required, this

simple model always winds up (and remains) in one

of the three states S, F or Inc.

4.2 Example Measurement Application

To illustrate an example of such a one goal

pathway, we consider an application to the medical

heart condition known as stroke. An input event is

the arrival of a patient with a possible stroke, at the

door of an emergency room. We employ the

formulation of “time lost is brain lost” of Toussaint

(2010) The goal is that the time from door to the

start of IV tPA (intravenous infusion blood clot

breakdown agent) be under 60 minutes. In Figure 5,

this means that the Activate port is triggered by

arrival of the patient at the door, the associated

question is whether IV tPA has been started, and the

time advance for the WG state is 60 minutes. Thus

the Yes input signifies infusion started before 60,

hence success. The input No arriving before 60

definitely indicates that the infusion will not be

started in time hence failure (this could happen if a

Figure 6: Health Care System Model.

subgoal such as taking a CAT scan failed due to

malfunction (see later discussion). Transition to the

Inc state indicates that information on the start of

the infusion did not arrive within the allowed time.

The general pathway atomic model has states

like WG for outputting questions and receiving

answers except that transition to an incomplete state

is optional. This will be illustrated in later

discussion Nevertheless, the requirement for finite

time is enforced so that eventually the

accomplishment of the main goal is known to be

true, false, or explicitly stated to be incomplete.

In Figure 6, the system-level model of

healthcare starts with a top down decomposition in

which a model of a Coordinated Care System (CCS)

is coupled with a Measurement System component

that monitors for arrival of patients with medical

conditions and other external events and evaluates

the resulting outcomes produced by the CCS. To

consider the Measurement System component we

need to discuss coupled pathway models.

4.3 Coupled Pathways Models

Coupling atomic pathway models enables us to

coordinate the behavior of multiple concurrent

pathways. For simplicity here, coupling will be

limited to activations by one pathway of one or

more others. DEVS closure under coupling will

assure that the resultant is a DEVS model. More

than that, we can show that the resultant is also

expressible as an atomic pathway model,

establishing closure under coupling when restricted

to the subset of DEVS defined as pathway models.

The following property is essential to such closure:

Finite Termination Property: For any pathway

model, there is a finite time T, such that the model

or all its components reach, and passivate, in any

one the three types of states: Success, Failed, or

Incomplete within time T after initialization.

The Appendix proves the Finite Termination

Property underlying closure of coupling.

4.4 Comparing Coordinated Care and

Traditional Clinical Pathways

Many of the features discussed above are common

to both coordinated care and clinical pathways

(Zeigler et al. 2014). However, coordinated care

pathways are focused on accomplishment of steps,

with associated accountability and payment

schemes. Consequently, they specify tests for

accomplishment and time bounds within which such

tests much be satisfied. While clinical pathways are

procedure oriented (i.e., tend towards increased

granularity in describing clinical processes), care

coordination pathways are more declarative (i.e.,

tend toward specification of goals and sub-goals

rather than procedures for achieving them.

Moreover, the underlying intent of Pathways are

quite different in two essential dimensions:

accountability and basis of payment. In a

protocol, accountability is not in a specific sense

taken into consideration. If the patient does not

show for follow-up appointments or the medication

Figure 7: Atomic Pathway for "time lost is brain lost".

isn’t being taking correctly, then the provider is not

held accountable as long as he/she followed the

protocol. This is not the case in a Pathway. The

Pathway is not considered complete until an

identified problem is successfully resolved.

Conversely, at some definitive point, a Pathway that

has not been successfully completed must be closed

in a documented fashion. Moreover, as indicated

above, coordinated care Pathways are associated

with payment for specific benchmarks along the

pathway with the highest payment provided for

successful outcomes at completion, thereby linking

payments to accomplishments.

5 PATHWAYS-BASED

MEASUREMENT

To apply the DEVS formalization of pathways to

measurement of goal achievement within deadlines,

we use the MS4 modeling and simulation

environment (ms4systems.com) to develop a

simulation model that enables design, exploration,

simulation and optimization. In Figure 6, the

Measurement System follows a patient starting with

his/her initial doctor visit or emergency room

appearance through interaction with an array of

services in the CCS. The Measurement System

tracks a) patients’ health status as they progress

through pathways in service care, and b)

accumulating the cost of care (tests, medications,

human care managers and providers) including cost

of readmissions to hospitals’ regular and emergency

departments.

5.1 DEVS Pathways Implementation

for the Measurement System

Porter’s Outcome Measurement Hierarchy (Porter

2010) provides a comprehensive basis for the

measurement system. Before discussing it in more

detail we prepare the groundwork by continuing the

discussion of the model of “time lost is brain lost”

above. We employ the DEVS pathway

representation for the Measurement System along

the lines of Porter’s Outcome Hierarchy design

approach. We define a comprehensive set of

outcome dimensions, and specific measures based

on the event-based experimental frame methods

implementable using DEVS.

Following the Pathways Coordination Model, this

will allow tracking patients through the full cycle

of care to accumulate actual costs of care (not how

they are charged, currently in arbitrary fashion).

The approach starts with the simplest, yet

meaningful, example of DEVS measurement

system. In the application to stroke, the activation

event is the arrival of a patient at the door and the

goal is that the infusion of the blood clot breakdown

agent, IV tPA start before 60 minutes have elapsed

since the patient’s arrival. The measurement system

must support detecting the events of patient arrival

and infusion occurrence and increment the count of

goal success if, and when, it does. The output is the

measure of success – percent of patients receiving

the injection in time. Figure 8 shows a coupled

model in which the atomic pathway model for “time

lost is brain lost” sends Success or Failed outputs to

an Accumulator model which counts patients and

percent of successful infusion of IV tPA before 60

minutes. Figure 8 shows the Accumulator as an

atomic model in the State Design graphical form

supported by MS4 Me. Note in this approach we

count Inc output as Failed. Zeigler et al .(2014)

discuss the inclusion of Incomplete in the counts of

outputs and the effects this has on the choice of

metrics of interest.

Figure 8: Measurement of Pathway Outcomes

Figure 9: Subgoals for the door-to-IV tPA process.

5.2 Alternative Architectures for

Sub-goal Tracking

Sub-goals for the door-to-IV tPA process are

formulated based on critical points in the process :in

which a CAT Scan is taken, read and interpreted.as

illustrated on the timeline in Figure 9, the subgoals

CT Scan and CT Read were established with

benchmarks of 25 and 45 minutes expiration from

arrival at the door (Toussaint, 2010) Note that the

Scan and Read measures are process measures not

outcomes of direct interest to the patient but that

can help the organization meet its overall goal.

Figure 7 also depicts an atomic pathway model with

states for CT Scan and CT Read subgoals where the

Scan must happen within 25 minutes and the Read

must take place within 20 minutes later for success.

Extending the model with a state for the goal of IV

tPA infusion would complete it.

The single atomic model, and its variations, is

one possible realization of measurement for the

door-to-IV tPA process. Coupled model alternatives

are sketched in Figure 10. Here the components are

pathway models for the individual subgoals and

goal: CT Scan, CT Read, and IV tPA , resp. These

components can be coupled in series or in parallel

as illustrated. The atomic model and coupled model

alternatives for realization of the measurement can

be considered as alternative architectures available

to explore as design options for implementation. In

the parallel case, each of the times is measured from

the arrival event at the door while in the sequential

case, these times are relative to the time of the

earlier stage completion. Thus the former allows

independent, less error prone, measurement while

the latter may be more natural to implement since it

conforms to the care delivery pathways.

Figure 10: Alternative Coupled Models for Multiple

Pathways.

Bradley et al. (2006) researched strategies to

reduce door-to-balloon time in Acute Myocardial

Infarction (AMI) including a time based goal of

expecting staff to arrive in the catheterization

laboratory within 20 minutes after being paged and

the catheterization laboratory to use real-time data

feedback. They found that despite the effectiveness

of these strategies, only a minority of hospitals

surveyed were using them- perhaps indicating the

need for further automation such as the current

approach can provide. The three alternative

architectures for what might be called Door-to-

critical-intervention can be represented as

specializations in a System Entity Structure (Zeigler

and Sarjoughian, 2013) and can be selected as

appropriate for different medical conditions. For

example, a heart attack (AMI) implementation

might use only a single atomic pathway model to

measure door-to-balloon times and survival rates. In

contrast, a stoke implementation might employ one

of the sequential or parallel alternative architectures

for its time-lost-is-brain-lost interventions.

5.3 Example: Coordinated HIV-AIDS

Care System Model

The continuity spectrum of HIV-AIDS intervention

spans HIV diagnosis, full engagement in care,

receipt of antiretroviral therapy, and achievement of

complete viral suppression (Figure 11). However,

Gardner.,et al., (2011) estimate that only 19% of

HIV-infected individuals in the United States have

been treated to the point where their virus is

Figure 11: HIV-AIDS Continuity of Care Pathway

Model.

undetectable. This occurs because achievement of

an undetectable viral load is dependent on

overcoming the barriers posed by patients “falling

through the cracks” in traversing each of the

sequential stages shown in Figure 11. The authors

conclude that recognition of the “pipeline” and

support for successful handoff of patients from

stage to stage is necessary to achieve a substantial

increase in successfully treated HIV population.

Figure 11 depicts the stages of care continuity

roughly assigned to both clinical and extra-clinical

domains and that they alternate between the two

domains (shown cycling from 1 to 4).

Coordination of the clinical and extra-clinical

domains is consistent with the multi-level

framework for coordination within and across

organizations (Gittell and Weiss, 2004). This

framework identifies key dimensions that can be

altered within and across organizations to enable or

improve communication and coordination. Such

dimensions include structure of the organization,

knowledge and technology employed, and

administrative operational processes and were

found to be effective in enhancing information

exchange, alignment of goals and roles, and

improved quality of relationships (Van Houtd,

2013). Here we consider the approach of

formulating the DEVs Pathways discussed above

for stages 1 and 3 to form an Integrated Practice

Unit. Also DEVS pathways are proposed for stages

2 and 4 which are similar to those of the Pathways

Community HUB. Using a DEVS coupled model,

the clinical domain pathways are interfaced to the

extra-clinical ones. The objective is that patients are

handed-off from one DEVS Pathway to the next

without being dropped from care. Such cross-

organization care pathways require sufficient

electronic health record system and health

information technology networking support to track

and monitor patients as they traverse the treatment

pipeline. Recall that this will require definition of

goals and subgoals along paths to complete cross-

system transactions, testing for achievement and

confirmation of pathway goals and subgoals,

tracking, and measuring progress of, patients along

the pathways they are following, and maintaining

accountability of compliance and adherence. The

implementation of such IT can then provide a

“dashboard” for viewing the overall disposition of

patients through the complete cycle of continuity of

care required for successful HIV-AIDS treatment.

6 ACTIVITY-BASED

CONTINUOUS

IMPROVEMENT

DEVS Pathways enable Time-Driven Activity-

based costing (Kaplan and Anderson, 2004) based

on pathway steps and their completion times. In this

regard, Muzy et al. (2013) identify three layers of

an adaptive system applicable to continuous

improvement of individual-based coordinated care

systems:

1. Time-Driven Activity-Based Costing using a

built-in measurement system.

2. Activity Evaluation and Storage: using the

built-in detection mechanisms of level 1, activity

can be measured as the fractional time that a

component contributes to the outcome. Correlating

contribution with outcome, a credit can be attributed

to components. Such a measure of performance of

components can be memorized in relation to the

experimental frame, or context, in which it

obtained.

3. Activity awareness: feedback of the activity-

outcome correlation to the inform the decision-

making process of which components or their

variations to apply overall or to the current problem.

Muzy and Zeigler (2014) describe a system that

implements these layers in an example simulation

experiment. Pathway coordination models lend

themselves to support critical features of such

learning systems. A pathway keeps track of

individuals’ traversal through the SoS As a DEVS

model, activity of a pathway over a time interval is

measured by the number of state transitions that

occurred in the interval (see (Zeigler et al. 2014) for

details.) The activity of the overall system is

estimated by the aggregation of all individual

pathway activities. When activity is aggregated over

all individuals that traversed a component, we get

an estimate of the component’s activity. These

measures can be sub-indexed by pathway to rank

the overall system activity from most active to least

active pathway, thereby providing insight into how

the system is being utilized. Further sub-indexing

by factors such as condition treated, patient

attributes, source of client referral, enable analysis

of the variation due to such factors. Moreover, since

pathways include outcome measurement they

enable correlation of activity and outcome for each

individual. Aggregation over individual traversals

of components yields component performance

outcome measures. Components or variants that do

not perform well in this measure are candidates for

replacement by other alternatives that can replace

them. Such activity-based outcome correlation and

feedback exhibits the continuous improvement

characteristic of an evidence-based learning

healthcare system advocated by Porter and others.

7 SUMMARY AND

CONCLUSIONS

The discussion presented can be summarized in the

following points:

 Health Care Reform is usefully viewed as a

System-of-Systems Engineering Problem.

 Coordination of healthcare was formulated as

an instance of Coordination Models defined as

client-oriented coordination of cross-system

transactions within a system-of-systems.

 Coordination Models were formalizing using

Pathways expressed in the DEVS modeling and

simulation formalism.

 The MS4 Modeling and Simulation

Environment based on DEVS supports design,

testing, and implementation of Coordination

Pathways in a systems engineering approach.

The President’s Council on Science and Technology

(PCAST ,2014) advocates increased use of systems

engineering in healthcare: “While there are

excellent examples, systems methods and tools are

still not used on a widespread basis through health

care.” The Systems of Systems Engineering

formalization and simulation modeling

methodology presented here will enable DEVS

Coordination Pathways to see wider use in re-

engineering of healthcare delivery systems.

Layering such models on top of the emerging health

data infrastructure will enable development of

system level metrics and analytics that will enable

healthcare to become a global learning system.

ACKNOWLEDGEMENTS

This material is based upon work supported by the

National Science Foundation under Grant Number

CMMI-1235364. Any opinions, findings, and

conclusions or recommendations expressed in this

material are those of the author(s) and do not

necessarily reflect the views of the National Science

Foundation. I acknowledge the essential intellectual

support of E. L. Carter, S. A. Redding, B. A. Leath, and

C. Russell who constituted the research team in this

project. I also acknowledge the essential technical

support of Dr. Chungman Seo, Mr. Wantae Kang,

and Mr. Keneth Duncan who were essential in

performing the modeling, simulation, and

implementation of pathways coordination

implementation. We greatly appreciate the help of

Drs. Mark and Sarah Redding, the originators of the

Communication HUB Pathways concept, in its

formalization.

REFERENCES

AHRQ, 2011. Connecting Those at Risk to Care: The

Quick Start Guide to Developing Community Care

Coordination Pathways.

http://www.innovations.ahrq.gov/guide/QuickstartGuide/C

ommHub_QuickStart.pdf.

AHRQ/NSF, 2009. “Industrial and Systems Engineering

and Healthcare: Critical Areas Workshop, September

21-22, 2009 workshop was co-sponsored by the

Agency for Healthcare Research and Quality

Augusto Vincent and Xiaolan Xie, 2014. A Modeling and

Simulation Framework for Health Care Systems,

IEEE Transactions on Systems, Man, and

Cybernetics: Systems, Volume:44 , Issue: 1, 30 – 46

Boardman, J., and Sauser, B. ,2006. System of Systems –

the meaning of "of". IEEE/SMC International

Conference on System of Systems Engineering. Los

Angeles, CA: IEEE.

Bradley Elizabeth H. et al. 2006 Strategies for Reducing

the Door-to-Balloon Time in Acute Myocardial

Infarction, N Engl J Med; 355:2308-2320November

30, 2006DOI: 10.1056/NEJMsa063117

Craig C and D, Whittington, 2011. J. Care Coordination

Model: Better Care at Lower Cost for People with

Multiple Health and Social Needs. IHI Innovation

Series white paper. Cambridge, Massachusetts:

Institute for Healthcare Improvement;. (Available on

www.IHI.org)

de Bleser L., R. Depreitere, K. de Waele, K. Vanhaecht,

J. Vlayen and W. Sermeus, 2007. "Defining

pathways," J Nurs Manag, vol. 14, pp. 553-63,

Gardner Edward M.,et al., 2011. The Spectrum of

Engagement in HIV Care and its Relevance to Test-

and-Treat Strategies for Prevention of HIV Infection

. Oxford Journals Medicine Clinical Infectious

Diseases Volume 52, Issue 6 Pp. 793-800

Gittell, JH and Weiss SJ. 2004.Coordination networks

within and across organisations: a multilevel

framework. J Management Studies, 41.(1):127–153.

Jamshidi, Mo, 2008. Systems of Systems – Innovations

for the 21st Century, Wiley,

Mesarovic, Mihajlo D. , D. Macko and Yasuhiko

Takahara 1970. Theory of Multi-level Hierarchical

Systems.. Academic Press.

Mittal, Saurabh and José L. Risco-Martín , 2012. DEVS

Net-Centric System of Systems Engineering with

DEVS Unified Process. CRC-Taylor & Francis

Series On System Of Systems Engineering,

Mittal, Saurabh, Bernard P. Zeigler, Jose L. Risco Martin,

Ferat Sahin and Mo Jamshidi, 2008. Modeling and

Simulation for Systems of Systems Engineering,

Systems of Systems -- Innovations for the 21st

Century, Edited by Mo Jamshidi, Wiley, , pp. 101-

149

Muzy Alexandre and Bernard P Zeigler, 2014. “Activity-

based Credit Assignment Heuristic for Simulation-

based Stochastic Search in a Hierarchical Model-

base of Systems”, submitted to IEEE Systems Jnl

Muzy, Alexandre et al., 2013. “Refounding of activity

concept? Towards a federative paradigm for

modeling and simulation”, in SIMULATION

Volume 89 Issue 2, February

Ören, Tuncer I. and B. P. Zeigler, 2012. “System

Theoretic Foundations of Modeling and

Simulation:A Historic Perspective and the Legacy of

A. Wayne Wymore”, SIMULATION/ September

vol. 88 no. 9 1033-1046

Ozcan, Yasar A.; Tànfani, Elena; and Testi Angela. 2011.

A simulation-based modeling framework to deal

with clinical pathways. Proceedings of the 2011

Winter Simulation Conference. S. Jain, R.R.

Creasey, J. Himmelspach, K.P. White, and M. Fu,

eds.

Porter, Michael E., 2010. “ Measuring Health Outcomes”,

New England Journal Of Medicine, December

Porter, Michael E. and Teisberg, Elizabeth Olmsted,

Redefining Health Care: Creating Value-based

Competition on Results, Harvard Business Review

Press; 1 edition (May 25, 2006)

PCAST ,2014. President’s Council of Advisors on

Science and Technology Report To The President

Better Health Care And Lower Costs: Accelerating

Improvement Through Systems Engineering,

http://www.whitehouse.gov/sites/default/files/micros

ites/ostp/PCAST/pcast_systems_engineering_in_hea

lthcare_-_may_2014.pdf (accessed June 9, 2014)

Seo, Chungman, Wontae Kang, Bernard P. Zeigler, and

Doohwan Kim, 2014. Expanding DEVS and SES

Applicability: Using M&S Kernels within IT

Systems. Symposium on Theory of Modeling &

Simulation – DEVS (TMS/DEVS) Spring Sim San

Diego, CA, USA

Shea, S.M. and C.E. Baker. 2013.Basketball Analytics:

Objective and Efficient Strategies for Understanding

How Teams Win. CreateSpace Independent

Publishing Platform,

Toussaint, John, 2010 ‘On the Mend: Revolutionizing

Healthcare to Save Lives and Transform the

Industry” 181 pages,: LEAN ENTERPRISE

INSTITUTE; 1 edition

Van Houdt S, Heyrman J, Vanhaecht K, Sermeus W, De

Lepeleire J, 2013. Care pathways across the primary-

hospital care continuum: using the multi-level

framework in explaining care coordination BMC

Health Services

Wickragemansighe N., et. al,. 2008. Health Care System

of Systems; in: Systems of Systems -- Innovations

for the 21st Century, Edited by Mo Jamshidi, Wiley,

, pp. 542-550

Wymore A. W., 1967. A Mathematical Theory of Systems

Engineering: The Elements, John Wiley & Sons, New

York, 1967.

Zeigler BP, 1989. "The DEVS Formalism: Event-Based

Control for Intelligent Systems," Proc. of the IEEE,

Vol. 77, No. 1, pp. 27-80.

Zeigler B P., Hessam S. Sarjoughian,2012.. Guide to

Modeling and Simulation of Systems of Systems,

Simulation Foundations, Methods and Applications,

Springer,.

Zeigler B.P., .G.Kim and H. Praehofer, 2000. Theory of

Modeling and Simulation: Integrating Discrete Event

and Continuous Complex Dynamic Systems, second

edition with co-authors T Academic Press, Boston,.

510 pages

Zeigler, B.P., and S. A. Redding., 2014. Question and

Answer Perspective on Care Coordination:

Formalization of Pathways for Standardization and

Certification, AHRQ Innovations Exchange Web

Site, http://www.innovations.ahrq.gov

Zeigler, B.P., Chungman Seo, And Doohwan Kim, 2013.

System Entity Structures for Suites of Simulation

Models, International Journal of Modeling,

Simulation, and Scientific Computing DOI:

10.1142/S1793962313400060, Volume 04, Issue 03,

September 2013

Zeigler, B.P., E. Carter, C. Seo, C. K Russell, and B. A.

Leath, 2012. Methodology and Modeling

Environment for Simulating National Health Care.

Autumn Simulation Multi-Conference

(AutumnSim'12), October 28-31, 2012.

Zeigler BP, Carter, EL Redding SA, Leath BA, Russell,

C, 2014. Care Coordination: Formalization of

Pathways for Standardization and Certification,

http://www.rockvilleinstitute.org/files/Care_Coordin

ation_Formalization_of_Pathways_for_Standardizati

on_and_Certification.pdf.(accessed April 9, 2014)

Summarized in Innovations Exchange Team, Zeigler

BP, Redding SA. Formalization of the Pathways

Model Facilitates Standards and Certification

http://www.innovations.ahrq.gov/content.aspx?id=40

97.(accessed April 9, 2014)

Zeigler BP, Carter, EL Redding SA, Leath BA, 2014.

Pathways Community HUB: A Model for

Coordination of Community Health Care, accepted

for publication in Population Health Management

APPENDIX

Preservation of Finite Termination Property:

Proof by induction:

Basis step: For atomic models, the property holds

by assumption that Requirement 3 holds.

Induction step: Assume the property holds for

coupled models with N components. Show it holds

for coupled models of N+1 components.

Let’s add a pathway model as a component to a

coupled pathway model of size N. By assumption,

the coupled model of size N can be considered as a

pathway atomic model in the sense of eventually

passivating in one of the allowable states. Consider

transferring of activation from one atomic pathway

model to another, This can only occur when the first

model reaches the Success state after outputting the

Activate. This output reaches a second pathway

model only if a coupling exists from the Activate

output port of the first to the Activate input port of

the second. We now show that the second model

will then become active if, and only if, it is in the

WA state when this Activate is received. The “if’

implication will be true since the model starts in the

passive WA state and cannot be disturbed from this

state by any input other than Activate. The “only if”

implication holds because no other state has such an

Activate input.

Thus, transfer of activation from one atomic to the

second through the Activate coupling, if this occurs

at all, will occur in a finite time after the first model

has been activated because of the basis step. Once

activated, an atomic pathway model will eventually

passivate in one of the allowable states after a finite

time, End of Proof.

BRIEF BIOGRAPHY

Bernard P Zeigler is Emeritus Professor of

Electrical and Computer Engineering at the

University of Arizona and Adjunct Research

Professor in the C4I Center at George Mason

University. He is internationally known for his

seminal contributions in modeling and simulation

theory and has published several books including

“Theory of Modeling and Simulation” and

“Modeling & Simulation-based Data Engineering:

Introducing Pragmatics into Ontologies for Net-

Centric Information Exchange“. He was named

Fellow of the IEEE for the Discrete Event System

Specification (DEVS) formalism that he invented in

1976. Among numerous positions held with the

Society for Modeling and Simulation International

(SCS) he served as President and was inducted into

its Hall of Fame. He is currently Chief Scientist

with RTSync Corp., a developer of the MS4

modeling and simulation software based on DEVS.

Zeigler’s research has been funded by a variety of

sponsors including National Science Foundation

(NSF), Defense Advanced Reseach Projects

Agency, US Air Force Research Laboratory among

others. . Currently, Zeigler is leading a project for

the NSF and the Agency for Healthcare Research

and Quality for the developing a simulation model

for the national healthcare system. For more

information see the Wikipedia entry on Bernard P

Zeigler.