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
Copyright
c
2014 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
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.