Healthcare Process Support: Achievements, Challenges, Current Research
Richard Lenz1, Mor Peleg2, Manfred Reichert3
1Friedrich-Alexander University Erlangen-Nuremberg, Germany
2University of Haifa, Haifa, Israel
3University of Ulm, Germany
ABSTRACT
Healthcare organizations are facing the challenge of delivering high-quality services to their patients at
affordable costs. To tackle this challenge, the Medical Informatics community targets at formalisms for
developing decision-support systems (DSSs) based on clinical guidelines. At the same time, business process
management (BPM) enables IT support for healthcare processes, e.g., based on workflow technology. By
integrating aspects from these two fields, promising perspectives for achieving better healthcare process support
arise. The perspectives and limitations of IT support for healthcare processes provided the focus of three
Workshops on Process-oriented Information Systems (ProHealth). These were held in conjunction with the
International Conference on Business Process Management in 2007-2009. The ProHealth workshops provided a
forum wherein challenges, paradigms, and tools for optimized process support in healthcare were
debated. Following the success of these workshops, this special issue on process support in healthcare provides
extended papers by research groups who contributed multiple times to the ProHealth workshop series. These
works address issues pertaining to healthcare process modeling, process-aware healthcare information system,
workflow management in healthcare, IT support for guideline implementation and medical decision support,
flexibility in healthcare processes, process interoperability in healthcare and healthcare standards, clinical
semantics of healthcare processes, healthcare process patterns, best practices for designing healthcare processes,
and healthcare process validation, verification, and evaluation.
Keywords: business process management, clinical guidelines, healthcare process support
INTRODUCTION
Healthcare organizations are facing the challenge
of delivering high-quality services to their patients
at affordable costs. Specialization of medical
disciplines, prolonged medical care for the ageing
population, increased costs for dealing with chronic
diseases, and the need for personalized healthcare
are prevalent trends in this information-intensive
domain. The emerging situation necessitates a
change in the way healthcare is delivered to
patients and healthcare processes are managed.
Enterprise-wide, process-oriented information
systems have been demanded by healthcare
institutions for over 20 years and terms like
“continuity of care” have even been discussed for
more than 50 years. Yet, healthcare organizations
are currently using a plethora of specialized, non-
standard information systems and continue
developing systems for specialized departments
that only consider internal processes. In particular,
many existing healthcare information systems are
still function- and data-centric, such as imaging,
drug order-entry, laboratory test result storage,
storage of diagnoses and progress notes in
electronic medical records, alerts and reminders,
and billing. Consequently, information systems and
decision-support systems (DSSs) managing patient
care processes are still scarcely developed. Such
patient care management systems are highly
complex and pose many challenges. For example,
they require availability of encoded data coming
from different sources and flexibility in deviating
from the implemented process at the discretion of
the user (e.g., physician). Further, they may involve
a team of clinical staff members that together take
care of a patient in a coordinated way.
The recent trend towards healthcare networks and
integrated care further increases the need to effec-
tively support interdisciplinary cooperation. Recent
studies discussing the preventability of adverse
events in medicine recommend the use of
information technology, since insufficient
communication and missing information have
turned out to be among the major factors
contributing to adverse events. Yet, there is still a
discrepancy between the potential and the actual
usage of IT in healthcare.
The three ProHealth workshops that were held
during 2007-2009 focused on the IT support of
healthcare processes. In particular, these workshops
brought together researchers from the Medical
Informatics and the BPM communities with the
vision that ideas from both fields and a mutual
understanding of relevant research issues will
create new insights and boost interdisciplinary
research. In the remainder of this introduction we
discuss what the two communities have achieved
so far and what top issues they have been
addressing. The latter will be done along with
suggestions of how to bring these issues together.
ISSUES ADDRESSED BY THE MEDICAL
INFORMATICS COMMUNITY
The Medical Informatics community has targeted
supporting patient care processes mainly by
developing electronic medical record (EMR)
systems (Hayrinen et al. 2008). These store patient
data and document patient encounters, thus keeping
a sustainable record of the patient's health state.
Further, DSSs have been developed (Peleg and Tu
2006) fostering decision-making and action
management. Special-purpose formalisms, known
as task-network models (Peleg et al. 2003), were
developed to specify evidence-based clinical
guidelines as representations that are computer-
interpretable and at the same time can be depicted
as a visual algorithm being easily understood by
medical experts. Unlike general-purpose process
models, the computer-interpretable guideline (CIG)
formalisms addressed issues that are important for
sharing guidelines among different implementing
institutions and for linking them to EMRs, focusing
more on the medical knowledge driving decision-
making rather than on organizational issues such as
scheduling and resource management. These issues
included (Peleg et al. 2003) use of standard clinical
terms, modeling of clinical concepts and medical
knowledge needed for decision-making, definitions
of clinical abstractions, patient information models
and expression languages used to specify temporal
clinical decision criteria, and representation of
different types of decision models, including if-
then-else rules, argumentation rules, and even
decision-theoretic models such as decision trees
and influence diagrams. Some of the guideline
models have also addressed other issues such as
representing the guideline's intentions (Shahar et al.
1998), allowing assessment of adherence to
intentions, and fitting into organizational
workflows (Quaglini et al. 2001; Tu et al. 2004).
The CIG formalisms are quite expressive and have
many benefits, yet using them to encode guidelines
is a long and difficult process. Several research
groups have focused on methods to ease,
standardize, and even partially automate guideline
encoding. Tools for marking-up guidelines have
been created (Karras et al. 2000; Peleg 2006;
Miksch 2007), design patterns for specifying
clinical guidelines and their components have been
developed (Miksch 2007; Peleg and Tu 2009), and
natural language processing has been used to parse
narrative guidelines in order to identify linguistic
patterns describing clinical actions (Serban et al.
2007). Researchers have also considered different
approaches for dividing and coordinating the
encoding task among people with different
expertise: knowledge engineers and clinical domain
experts (Peleg et al. 2008; Shalom et al. 2008).
Since clinical guidelines provide recommendations
for care and not a strict inflexible assembly line,
they are modeled as flexible care plans that can be
executed according to physicians' discretion, who
can deviate from the original care plan. Allowing
such deviations (Quaglini et al. 2000) yet
controlling and managing their scope, in order to
prevent medical errors, is an active line of research
as well as assessing clinicians' compliance to
clinical guidelines (as specified by a formal
guideline model) (Advani et al. 1998; Micieli et al.
2002; Advani et al. 2003).
The focus of the Medical Informatics community is
complementary to that of the BPM community,
allowing synergism. As suggested in (Terenziani
2009), an integration of the approaches used for
healthcare process management by the two
communities could potentially be achieved using a
hybrid approach in which a computer-interpretable
guideline approach is used to focus on “physician-
oriented” issues, a workflow approach is used to
cope with the related “business-oriented” issues,
and the integration of them is obtained at the
underlying semantic level, where also general
inferential mechanisms operate.
ISSUES ADDRESSED BY THE BPM
COMMUNITY
Historically speaking, business process support has
been a major driver for enterprise information
systems for a long time. The overall goal is to
overcome the drawbacks of functional over-
specialization and lack of overall process control.
Technology response to this business demand was
met with a suite of technologies ranging from
office automation to workflow systems to BPM
technology.
Just as DBMS provide a means of abstracting
application logic from physical data aspects,
workflow management systems (WfMS) separate
coordinative process logic from application code
(Leymann and Roller 2000). Although workflow
technology has delivered a great deal of
productivity improvements, it has been mainly
designed for the support of pre-specified and
repetitive business processes requiring a basic level
of coordination between human performers and
some application services. More recently BPM has
been used as broader term to reflect the fact that a
business process may or may not involve human
participants, and often crosses organizational
boundaries.
Currently, there is a widespread interest in BPM
technologies, especially in the light of emerging
paradigms surrounding service-oriented computing
and its application to dynamic service orchestra-
tions and choreographies. In this context, the notion
of PAIS (Process Aware Information System)
provides a guiding framework to understand and
deliberate on the above developments (Dumas et al.
2005). As fundamental characteristic, a PAIS
provides the basic means to separate process logic
from application code. Furthermore, challenges,
features and limitations of existing PAISs can be
discussed along the phases of the process lifecycle
(Weber et al. 2009; Weber et al. 2009); e.g.,
process design, implementation & configuration,
enactment, monitoring & diagnosis, and evolution.
At process design time the process logic has to be
explicitly defined based on the constructs provided
by a process modeling language. In this context, a
variety of workflow patterns (e.g., control and data
patterns, resource patterns, time patterns) are
suggested, enabling the comparison and evaluation
of existing modeling languages (van der Aalst et al.
2003; Lanz et al. 2010). Other work, in turn, targets
involvement of end users in the design process by
increasing model quality and understandability
(Mendling et al. 2007; Weber et al. 2011). At
process run-time a PAIS orchestrates the processes
according to the defined logic (i.e., process model)
and coordinates corresponding applications and
other resources. Examples of PAIS-enabling
technologies include WfMSs like WebSphere
Process Server (Kloppmann et al. 2008), ADEPT2
(Dadam and Reichert 2009), AristaFlow (Reichert
et al. 2009), and YAWL (van der Aalst and ter
Hofstede 2005) as well as case handling
frameworks like FLOWer (van der Aalst et al.
2003) and PHILHarmonic Flows (Kunzle and
Reichert 2011).
In spite of several success stories on the uptake of
PAISs and the growing process orientation of
enterprises, BPM and related technologies have not
had the widespread adoption that was expected. A
major reason for this is the limited process
flexibility offered by existing PAISs, which inhibits
the ability of an organization to respond to business
changes and exceptional situations in an agile way
(Dumas et al. 2005; Weber et al. 2009). To deal
with exceptions, uncertainty, and evolving
processes, it is widely recognized that a PAIS
needs to provide run-time flexibility (Reichert et al.
2009). This can either be achieved through
dynamic structural process changes or by
supporting loosely specified process models, which
can be refined during run-time according to pre-
defined criteria and rules. To address this need
paradigms like adaptive processes (Reichert and
Dadam 1998), case handling (van der Aalst et al.
2003; van der Aalst and ter Hofstede 2005), and
declarative processes (Pesic et al. 2007) have
emerged. Generally, they can be characterized
along three fundamental requirements, namely
support for flexibility, adaptation, and evolution:
Flexibility represents the ability of the
implemented process to execute on the basis of
a loosely or partially specified model which is
completed at run-time and may be unique to
each process instance (i.e., case). Due to the
high number of choices, not all of which can
be anticipated and hence pre-specified,
frameworks like DECLARE (van der Aalst et
al. 2003; Pesic et al. 2007), Alaska (Zugal et
al. 2011), and PocketsOfFlexibility (Sadiq et
al. 2005) allow defining process models in a
more relaxed manner; the model can be
defined in a way that allows individual
instances to determine their own (unique)
processes. In particular, declarative approaches
allow for loosely-specified process models by
following a constraint-based approach. While
pre-specified process models define exactly
how the overall task has to be accomplished,
constraint-based process models focus on what
should be done by describing the set of
activities that may be performed as well as the
constraints prohibiting undesired process
behavior. Generally, loosely specified models
raise several challenges including the flexible
configuration of process models at design time
(Hallerbach et al. 2010) or their constraint-
based definition during runtime (van der Aalst
et al. 2003; Pesic et al. 2007).
Adaptation represents the ability of the
implemented processes to cope with
exceptional circumstances. On the one hand,
existing PAISs like YAWL provide support for
the handling of expected exceptions, which can
be anticipated and thus be captured in the
process model (Russel et al. 2006).
Alternatively, adaptive PAISs like ADEPT2
also cover the handling of unanticipated
exceptions, which are usually addressed
through structural ad-hoc changes of single
process instances (e.g., to add, delete or move
process steps during process execution)
(Reichert and Dadam 1998; Reichert et al.
2009; Weber et al. 2009). Clearly, such
dynamic process adaptations necessitate a
comprehensive framework ensuring
correctness and robustness of the PAIS in the
context of ad-hoc changes as well.
Evolution represents the ability of a process
implemented in a PAIS to change when the
business process evolves (e.g., due to legal
changes or process optimizations). The
assumption is that the processes have pre-
specified models, and a change causes these
models to be modified. The biggest challenge
is the handling of the potentially large number
of long-running process instances, which were
initiated based on the old model but are
required to comply with the new specification
from now on. Approaches like WASA2,
ADEPT2 and WIDE allow process engineers
to migrate such process instances to the new
model version, while ensuring PAIS
robustness and process consistency (see
(Rinderle et al. 2004) for these approaches).
In practice there often exists a significant gap
between what is prescribed and what actually
happens. Generally, a PAIS records the actual
execution behavior of a collection of process
instances in an execution log. Furthermore, in
adaptive and flexible PAISs, deviations from the
pre-specified model can be recorded in change
logs.
In this context process mining strives to deliver a
concise assessment of the organizational reality by
mining these logs. Generally, there exist different
classes of process mining techniques. Process
discovery algorithms analyze execution logs and
derive process models from them reflecting the
actual process behavior best (van der Aalst et al.
2007). Conformance testing (Rozinat and van der
Aalst 2008) analyzes and measures discrepancies
between the original model of a process and the
actual execution of its instances (as recorded in
execution logs). Log-based verification (van der
Aalst et al. 2005), in turn, checks the log for
conformance with desired or undesired properties;
e.g., process instance compliance with corporate
guidelines or global regulations. Furthermore,
change mining techniques (Gunther et al. 2008) do
not only consider the execution logs of process
instances, but additionally analyze the structural
changes applied during process execution; i.e., they
allow visualizing and analyzing dynamic deviations
from the original process model. Finally, process
variants mining allows discovering an optimal
reference process model being “close” to a given
collection of process variants (e.g., process
instances derived from the same model, but
structurally differing due to ad-hoc changes applied
to them) (Li et al. 2010).
GAPS AND CHALLENGES
To explain the challenges in supporting processes
in healthcare it is important to distinguish the
patient-specific medical treatment process from the
organizational process that generally coordinates
the cooperation between various process
participants and organizational units within a
healthcare institution (Lenz and M Reichert 2007).
Medical informatics research starts from the
medical tasks and problems of supporting daily
work of physicians. This has improved the
understanding of how healthcare processes actually
work and how complex they are. A number of
techniques supporting doctors in clinical decisions
have been developed, and some of these, such as
alerts and reminders, have already been proven to
be effective in preventing adverse events when
being used properly (Shea et al. 1996; Del-Fiol et
al. 2008). Furthermore, standards for data
interchange have been developed in order to
improve the integration of heterogeneous systems
and thereby improve the basis for an optimized
support of organizational processes. It was
demonstrated that the usage of order entry and
result reporting could improve the quality of
healthcare (Overhage et al. 1997), however,
modern workflow management technology which
promises more transparency and flexibility in
process management has not found its way into
hospitals yet. The BPM community started with
evaluating these technologies and their weaknesses,
e.g., how can it be improved and adapted to various
real-world requirements and, in particular, how can
it be utilized to support healthcare processes.
Researchers from the BPM community have
developed numerous techniques to model and
verify business processes as well as workflow
management solutions that are adaptable to a broad
spectrum of processes.
Both communities have addressed important
aspects of supporting healthcare processes and the
contributions of the communities seem to be quite
complementary. CIG modeling languages often do
not address important managerial activities such as
linking these guidelines to order-entry systems,
scheduling visits, and so forth. On the other hand,
BPM approaches often do not address complex
decision criteria that include medical abstractions
and temporal expressions bridging the gap between
the medical knowledge abstractions used in clinical
guidelines and the actual raw data stored in EMRs.
Yet, we still have not seen complete integrations of
BPM solutions with computer-interpretable clinical
guideline solutions, except in the case of decision-
support systems developed in the Guide guideline
modeling language (Quaglini et al. 2001).
Thus, there is still a gap between domain- and
technology-driven approaches. In addition to the
aspects addressed so far, many problems have been
recognized, but not been solved in an acceptable
way yet, and some have even not been understood
well enough to suggest workable solutions. Some
of these problem areas can be briefly summarized
as follows:
Integration of heterogeneous systems. Both
WfMSs and DSSs suffer from the fact that
operational data in healthcare processes is
typically stored in heterogeneous and often
autonomous IT systems. Standards for data
interchange help to decrease the semantic
heterogeneity of data to some degree. Further,
approaches for matching clinical abstractions
found in clinical guidelines to patient data
found in EMRs have been suggested (Peleg et
al. 2008; German et al. 2009). Finally, even
commercial solutions for presenting data from
various EHRs in a harmonized way
(www.dbmotion.com) exist. However, still
support for cooperation across systems has not
fully been provided and problems such as
access control have not been sufficiently
addressed.
Social issues. Adequately embedding IT
support into routine work practices is one of
the greatest challenges. Functionality alone
will be not enough if it is not accepted or
cannot be adequately used for actually
optimizing the healthcare process. Medical
informatics research has identified numerous
generally applicable rules of thumb for change
management. Heeks proposed the design-
reality gap model as a tool that helps assessing
the risks of IT projects in healthcare (Heeks
2006). Factors influencing success and failure
of healthcare IT projects have been identified (
e.g., (Sauer 1993; Lorenzi and Riley 2000; Ash
et al. 2003)) and, more specifically, general
recommendations on how to successfully
implement medical guidelines have been given
(e.g., (Bates et al. 2003)). However, it is
unclear how BPM techniques can contribute to
improve process alignment in healthcare, i.e.,
the alignment of IT functionality and
healthcare processes, or in other words the
adequate embedding of IT services into routine
work practice in order to maximize its benefit.
Evolving processes. The problem of process
alignment is aggravated by continuously
evolving processes due to advances in clinical
medicine (e.g., new medications, new
laboratory tests, and new evidence from
clinical trials) as well as process optimization.
The challenge is to support continuous
learning and ongoing change by IT systems.
While it is the responsibility of medical
societies to evolve the non-organizational
specific healthcare process model, it is the
responsibility of the local organization to
introduce these changes into their integrated IT
systems. However, recommendations in
clinical guidelines typically do not explain
how to implement them in a given
organizational setting.
Process tracking. Another important unsolved
problem is how to deal with real-world process
that deviate from process documentation.
Process mining techniques have helped to
learn how real process instances behave by
analyzing event logs. However, unlike
production processes where progress can be
precisely monitored, healthcare processes may
contain undocumented process change events.
This makes process monitoring and tracking
particularly challenging and it requires process
participants to be aware of the potential
discrepancy between the documented process
and the real process. Note, that – though
related – this problem area covers more than
guideline compliance, e.g., (Advani et al.
1998); guideline compliance is the degree of
how well a process instance matches with a
given guideline. The problem described here is
different: the documentation of a process
instance in the IT system might deviate from
what really happened – it might be incomplete
or even incorrect (Lenz, Blaswer et al. 2007).
These open issues are unlikely to be solved in the
near future, but they serve as a motivation to
continue with interdisciplinary research, and this is
at the core of the ProHealth workshops.
CURRENT RESEARCH FROM PROHEALTH
Current research initiated by the collaboration
between the two communities that resulted in the
ProHealth workshops has addressed a variety of
topics. Selected topics from the three PROhealth
workshops are summarized in the following.
Workflow-pattern based analysis of CIG
formalisms. The aim of this line of research is to
study existing CIG formalisms with the aim of
standardizing and improving their control flow
semantics based on the vast experience
accumulated by research on workflow languages.
Because much of the resistance by the medical
informatics community to use workflow
technology has stemmed from the myth that
workflow languages are not flexible enough to
model clinical processes, Mulyar et al. (Mulyar et
al. 2007) analyzed four guideline formalisms in
terms of their support of workflow control-flow
patterns. They have shown that, in fact, the
guideline formalisms support only a limited set of
control-flow patterns. Of 43 workflow patterns,
PROforma supported 23 patterns; Asbru supported
20, GLIF 17, and EON only 11. There was no
support of the multi instance activity pattern and
the semantics of Synchronizing Merge was
imprecise. In another research, Mulyar et al.
(Mulyar et al. 2007) have shown that declarative
approaches are more flexible than existing
guideline formalisms. They have demonstrated how
the templates of the CIGDec declarative
specification language enable the control-flow
constraints of typical healthcare scenarios.
Inspired by the papers described above, Grando et
al. (Grando et al. 2008) examined the question of
which flexible control-flow patterns are supported
by the PROforma guideline modeling language.
The authors defined a mapping from the PROforma
language to Colored Petri Nets and utilized it to
construct formal proofs that PROforma is capable
of expressing a standardized workflow pattern.
Flexible IT support for healthcare processes.
One of the fundamental challenges discussed in all
ProHealth workshops concerns process flexibility.
Physicians often have to decide which diagnostics
or therapies are necessary or may be dangerous due
to contraindications and treatment-typical pro-
blems. Generally, decisions about the next steps
have to be made during the treatment process by
interpreting patient-specific data according to
medical knowledge and considering the current
state of the patient. As opposed to organizational
processes (e.g., order handling) such knowledge-
intensive processes usually cannot be fully pre-
specified and automated. For example, there are
clinical guidelines recommending evidence-based
compilations of care processes. However, such care
processes cannot account for all possible treatment
cases and therefore, the PAIS that executes the
guideline-based care processes must allow for
deviations. Generally, physicians are not supposed
to obey any step-by-step process, but need to
provide the best possible treatment for their
patients taking the given situation into account.
Mans et al. identified flexibility requirements to be
met in order to adequately support the various
kinds of healthcare processes (Mans et al. 2008).
The authors have shown that several process
support paradigms are needed to adequately cope
with this challenge (see (Mulyar et al. 2007) for
similar considerations emphasizing the need for
supporting both procedural and declarative
paradigms in connection with the modeling of
clinical guidelines).
PAISs relying on pre-specified process models,
which are the predominant paradigm for modeling
and executing processes, have been applied to
healthcare processes for more than a decade. For
them a variety of techniques for accommodating
the need for flexibility, adaptation and evolution is
provided. In the context of ProHealth, van Hee et
al. introduced adaptive workflow nets for the
flexible modeling of care processes (van Hee et al.
2008). Reijers et al added a methodology for
capturing healthcare processes based on a number
of workflow and flexibility patterns (Reijers et al.
2009). While these works focus on a particular
phase of the process lifecycle, a few approaches
stemming from the BPM community enable full
process lifecycle support. As example, consider the
ProCycle framework (Weber et al. 2009), which
enables integrated support of all phases of the pro-
cess lifecycle ranging from modeling to enactment
to ad-hoc adaptations to process learning to process
evolution. Finally, assistance for end users in
exceptional situations is provided; e.g., by allowing
them to reuse previously applied ad-hoc changes
when a similar problem context is given.
During the last years, declarative approaches have
been applied to clinical guidelines (Lyng et al.
2008; van der Aalst et al. 2009). They suggest a
fundamentally different way of describing pro-
cesses being promising for the support of dynamic
patient treatment processes. For example, Declare
and Alaska enable loosely-specified process
models by allowing users to defer modeling deci-
sions to run-time. Potential advantages include the
absence of over-specification and the provision of
more maneuvering room for end users. However,
more and more it is recognized that knowledge-
intensive healthcare processes cannot always be
straightjacketed into activities. Prescribing an
activity-centric process model for them would lead
to a "contradiction between the way processes can
be modeled and the preferred work practice" (Sadiq
et al. 2005). Instead object-awareness is required;
i.e., full integration of processes with application
data consisting of object types and object relations.
In accordance to a (patient) data model comprising
object types and object relations, therefore, the
modelling and execution of patient-related
processes can be based on two levels of granularity:
object behaviour and object interactions (see
(Kunzle and Reichert 2011) for a respective
framework from the BPM community). Recently,
Neumann and Lenz have picked up this metaphor.
With alpha-flow they suggest a document-based
approach to the flexible support of inter-
departmental healthcare processes (Neumann and
Lenz 2009; Neumann and Lenz 2009).
Verification and testing of healthcare process
models. The BPM community has been working
for many years on methods for verifying and
testing business processes. Developing methods for
the healthcare domain, Imam and MacCaull (Imam
and MacCaull 2008) created a multi-threaded
model checker to reason about timed processes in
careflows sensitive to patient preferences and care
team goals, using a temporal logic extended with
modalities of beliefs, desires and intentions. In
another paper by that group (Miller and MacCaull
2009), they have developed a multi-valued logic
based system that allows merging two inconsistent
terminologies.
Osterweil, Clarke &Avrunin (Osterweil et al. 2009)
developed the Little-JIL process definition
language and an integrated collection of tools
supporting the precise definition, analysis, and
execution of processes that coordinate the actions
of humans, automated devices, and software
systems for the delivery of healthcare. It is intended
to support the continuous improvement of the
healthcare delivery processes. Another approach is
proposed in (Mans et al. 2009), where the same
model is used for specifying, developing, testing
and validating the operational performance of a
new system. This approach has been applied to a
schedule-based workflow system developed for the
AMC hospital in Amsterdam.
Coping with semantic heterogeneity in
autonomous systems. An important problem
complicating process support in healthcare is the
semantic heterogeneity of autonomous systems
Neumann and Lenz (Lenz, Beyer et al. 2007)
propose a document-based approach to support
cooperation in healthcare networks. The basic idea
is to use self-describing electronic documents as
the unit for information interchange. By including
process related metadata into independent
electronic documents, inter-institutional processes
can be supported without the need to closely
interconnect pre-existing IT-systems. An important
aspect of this approach is the strict separation of
coordinating activities from document contents. By
separating these aspects semantic interoperability
can be addressed independent of basic coordination
tasks, thereby enabling cooperation without the
need for prior integration of existing IT systems.
Process mining and goal-based process learning.
Inspired by the BPM community, research related
to the goals of clinical processes was once again
brought to the focus. Early work by Shahar et al.
(Shahar et al. 1996; Shahar et al. 1998) focused on
intention-based specification of clinical guidelines,
where the intentions of guideline plans and their
refined lower-level actions were specified as
temporal patterns, and critiquing of guideline
application based on compliance to the guideline's
intentions.
Current research drew ideas and methods from the
vast amount of work on process mining done by the
BPM community. Mans et al. (Mans et al. 2008)
used process mining to discover non-compliance
with a stroke guideline and reasons for it. This
helped in reconstructing chains of responsibilities
concurring to produce errors in a complex patient’s
pathway, learning how to improve clinical
guidelines (Quaglini 2008).
Ghattas, Soffer, and Peleg (Ghattas et al. 2007;
Peleg et al. 2007; Ghattas et al. 2009) used process
mining at the semantic level to improve healthcare
processes. In this approach, healthcare process
instance data is used to learn the best path needed
in order to achieve desired outcomes for patients
with different contextual characteristics. Using a
case study of a urinary tract infection care process,
they used machine-learning techniques to find the
important patient groups, based on similarity of
process paths and outcome. They then used a
decision-tree learning algorithm to discover
contextual data items that could predict the
partition into these patient groups; From the
decision tree, a semantic definition of the context
groups was discovered.
Goal-based approach for exception-handling.
The focus that the ProHealth workshop has given to
goals in clinical processes inspired Grando and
colleagues (Grando et al. 2010; Grando et al. 2010)
to develop a goal-based framework that can be
used to monitor, detect, and handle exceptions
occurring during normal CIG execution. This, in
turn, can potentially prevent them from evolving
into medical errors. This framework (Grando et al.
2010) allows specifying the goals of a clinical
guideline and linking them with recommended
tasks that could satisfy these goals. Exceptions are
linked with goals that manage them, which can be
realized by tasks or plans. To achieve a link
between the tasks, plans, goals, monitored effects,
and exceptions, the definition of goals and
exceptions is state-based. The goal-based approach
for exception handling was demonstrated in the
domain of hypertension management. In (Grando et
al. 2010), the authors extended this framework to
deal with exceptions arising from miscommunica-
tion that can happen when an actor in an
organization assigns an action to another actor. To
support this, the goal-based exception handling
framework has been extended to formally specify
the transfer of responsibility and accountability
when tasks are delegated in healthcare teams.
DISCUSSION
The ProHealth workshop series has opened an on-
going interdisciplinary and fruitful discussion on
process support in healthcare. Yet, considering the
different backgrounds of the participants, it takes
effort to build a common ground that can serve as a
basis for improving mutual understanding. The
ongoing process of building common ground is
supported by having each submitted paper re-
viewed by three reviewers – typically with different
backgrounds. It is further supported by inviting
experienced speakers presenting the condensed
experience gathered over years in their particular
research communities. And finally, an important
aspect is to leave enough room for discussion.
First results from the ProHealth community-
building efforts have been summarized in this
editorial paper. Even if details have not been
elaborated yet, the general feasibility of applying
BPM technology to healthcare has been shown:
Mapping guidelines to formal models enables
simulation, validation, execution, and verification
as well as analyzing the flexibility of control-flow.
Goal-oriented monitoring, for example, can help to
prevent critical situations that may turn into
medical errors. The kinds of flexibility needed to
adequately support healthcare processes are better
understood. Process mining can give valuable
feedback and support organizational learning. Yet,
the fact that real-world processes are typically not
completely recorded in event logs raises the need to
cope with incomplete and imprecise information.
Integration of heterogeneous systems is needed to
enable cross-organizational cooperation, and
semantic integration is the hardest part to achieve.
It could be demonstrated that separating semantic
integration from basic coordination tasks is helpful
to overcome communication barriers and to enable
continuous improvement of cooperation.
Such results encourage us to continue the dialog
and try to learn from each other. The contributions
have shown that the participants are actually not
only members of two communities. Particularly the
discussion with Lee Osterweil and Barbara Paech,
who organize a similar Workshop in conjunction
with the International Conference on Software
Engineering showed that other disciplines can also
contribute to achieve valuable improvements for a
multi-faceted problem area. In order to broaden the
interdisciplinary dialog Lee Osterweil has agreed to
give the invited talk at ProHealth 2011.
REFERENCES
W.M.P. van der Aalst, M. Weske, and D.
Grünbauer: Case handling: a new paradigm for
business process support. Data & Knowledge
Engineering, 53(2): 129-162, 2003.
W.M.P. van der Aalst, H.T. de Beer, and B.F. van
Dongen: Process mining and verification of
properties: an approach based on temporal logic.
In: Proc. CoopIS'05, pp. 130-147, 2005.
W.M.P. van der Aalst, A.H.M. ter Hofstede, B.
Kiepuszewski, and A. Barros: Workflow patterns.
Distributed & Parallel Databases, 14(1): 5-51,
2003.
W.M.P. van der Aalst and A.H.M ter Hofstede:
YAWL: Yet another workflow language.
Information Systems, 30(4): 245-75, 2005.
W.M.P. van der Aalst, M. Pesic, and H.
Schonenberg: Declarative workflows: balancing
between flexibility and support. Computer Science
- Research & Development, 23(2): 99-113, 2009.
W.M.P. van der Aalst, H.A. Reijers, A. Weijters,
B.F. van Dongen, A.K. de Medeiros, M. Song, and
H. Verbeek: Business process mining: an industrial
application. Inf. Systems, 32(1): 713-32, 2007.
A. Advani, M. Goldstein, Y. Shahar, and M.A.
Musen: Developing quality indicators and auditing
protocols from formal guideline models:
knowledge representation and transformations. In:
Proc. AMIA Symposium, pp. 11-15, 2003.
A. Advani, K. Lo, and Y. Shahar: Intention-based
critiquing of guideline-oriented medical care. In:
Proc. AMIA Symposium, pp. 483-487, 1998.
J. Ash, Z. Stavri, and G.J. Kuperman: A consensus
statement on considerations for a successful CPOE
implementation. J Am Med Inform Association,
10(3): 229-234, 2003.
D.W. Bates, G.J. Kuperman, S. Wang, T. Gandhi,
A. Kittler, L. Volk, C. Spurr, R. Khorasani, M.
Tanasijevic, and B. Middleton: Ten command-
ments for effective clinical decision support:
making the practice of evidence-based medicine a
reality. J Am Med Inform Association, 10(6): 523-
530, 2003.
P. Dadam and M. Reichert: The ADEPT project: a
decade of research and development for robust and
flexible process support - challenges and
achievements. Computer Science - Research &
Development, 23(2): 81-97, 2009.
G. Del-Fiol, P.J. Haug, J.J. Cimino, S.P. Narus, C.
Norlin, and J.A. Mitchell: Effectiveness of topic-
specific infobuttons: a randomized controlled trial.
J Am Med Inform Assoc, 15(6): 752-59, 2008.
M. Dumas, A.H.M. ter Hofstede, and W.M.P. van
der Aalst: Process-aware information systems,
Wiley Publishers, 2005.
E. German, A. Leibowitz, and Y. Shahar: An
architecture for linking medical decision-support
applications to clinical databases and its evaluation.
J Biomed Inform, 42(2): 203-18, 2009.
J. Ghattas, M. Peleg, P. Soffer, and Y. Denekamp:
Learning the context of a clinical process. In: Proc.
Business Process Management Workshops, LNBIP
43, pp. 545-556, 2009.
J. Ghattas, P. Soffer, and M. Peleg: Learning
business process models: a case study. In: Proc.
Business Process Management Workshops, LNCS
4928, pp. 383-394, 2007.
A.M. Grando, M. Peleg, and D. Glasspool: A goal-
oriented framework for specifying clinical
guidelines and handling medical errors. J of
Biomedical Informatics, 1(2): 287-299, 2010.
A.M. Grando, M. Peleg, and D. Glasspool: Goal-
based design pattern for delegation of work in
health care teams. In: Proc. Medinfo'10, pp. 299-
303, 2010.
A.M. Grando, D. Glasspool, and J. Fox: Petri Nets
as a formalism for comparing expressiveness of
workflow-based clinical guideline languages. In:
Proc. Business Process Management Workshops,
LNBIP 17, pp. 348-360, 2008.
C.W. Günther, S. Rinderle-Ma, M. Reichert,
W.M.P. van der Aalst, and J. Recker: Using
process mining to learn from process changes in
evolutionary systems. Int J of Business Process
Integration and Management, 3(1): 61-78, 2008.
A. Hallerbach, T. Bauer, and M. Reichert:
Capturing variability in business process models:
the Provop approach. Journal of Software
Maintenance and Evolution: Research and
Practice, 22(6-7): 519-546, 2010.
K. Hayrinen, K. Saranto, and P. Nykanen:
Definition, structure, content, use and impacts of
electronic health records: a review of the research
literature. Intl J Med Inform, 77(5): 291-304, 2008.
K. van Hee, H. Schonenberg, A. Serebrenik, N.
Sidorova, and J.M. van der Werf: Adaptive
workflows for healthcare information systems. In:
Proc. Business Process Management Workshops,
LNBIP 17, 359-370, 2008.
R. Heeks: Health information systems: failure,
success and improvisation. Int J Medical Inform,
75: 125-37, 2006.
F.T. Imam, and W. MacCaull: Integrating
healthcare ontologies: inconsistency tolerance and
case study. In: Proc. Business Process Mgmt.
Workshops, LNBIP 17, 373-384, 2008.
B.T. Karras, S.D. Nath, and R.N. Shiffman: A
preliminary evaluation of guideline content mark-
up using GEM: an XML guideline elements model.
In: Proc. AMIA Symp, pp. 413-417, 2000.
M. Kloppmann, D. Knig, F. Leymann, G. Pfau, and
D. Roller: Business process choreography in
WebSphere: combining the power of BPEL and
J2EE. IBM Systems Journal, 43(2): 270-96, 2008.
V. Künzle and M. Reichert: PHILharmonicFlows:
towards a framework for object-aware process
management. Journal of Software Maintenance and
Evolution: Research & Practice, 23(4): 205-244,
2011.
A. Lanz, B. Weber, and M. Reichert: Workflow
time patterns for process-aware information
systems. In: Proc. BPMDS, LNBIB 50, 94-107,
2010.
R. Lenz, M. Beyer, and K. Kuhn: Semantic
integration in healthcare networks. Int'l J Med
Inform, 76(2-3): 201-207, 2007.
R. Lenz, R. Blaser, M. Beyer, O. Heger, C. Biber,
M. Baumlein, and M. Schnabel: IT support for
clinical pathways: lessons learned. Int'l J Med
Inform, 76(3): 397-402, 2007.
R. Lenz and M. Reichert: IT support for healthcare
processes - premises, challenges, perspectives.
Data & Knowledge Eng, 61(1): 39-58, 2007.
F. Leymann and D. Roller: Production Workflow:
Concepts and Techniques, Prentice Hall, 2000.
C. Li, M. Reichert, and A. Wombacher: The
MinAdept clustering approach for discovering
reference process models out of process variants.
Int'l J of Cooperative Information Systems, 19(3-4):
159-203, 2010.
N.M. Lorenzi and R.T. Riley: Managing change: an
overview. J Am Med Inform Assoc, 7(2): 116-24,
2000.
K.M. Lyng, T. Hildebrandt, and R.R. Mukkamala:
From paper based clinical practice guidelines to
declarative workflow management. In: Proc.
Business Process Management Workshops, LNBIP
17, pp. 336-347, 2008.
R.S. Mans, H. Schonenberg, G. Leonardi, S.
Panzarasa, A. Cavallini, S. Quaglini, and W.M.P.
van der Aalst: Process mining techniques: an
application to stroke care. Stud Health Technol
Inform, 136: 573-578, 2008.
R.S. Mans, W.M.P. van der Aalst, N.C. Russell,
and P.J.M. Bakker: Flexibility schemes for
workflow management systems. In: Proc. Business
Process Management Workshops, LNBIP 17, pp.
361-372, 2008.
R.S. Mans, W.M.P. van der Aalst, N.C. Russell,
P.J.M. Bakker, and A.J. Moleman: Process-aware
information system development for the healthcare
domain: consistency, reliability, and effectiveness.
In: Proc. Business Process Management Work-
shops, LNBIP 43, 635-646, 2009.
J. Mendling, H.A. Reijers, and J. Cardoso: What
makes process models understandable? In: Proc.
BPM'07, LNCS 4714, 48-63, 2007.
G. Micieli, A. Cavallini, and S. Quaglini: Guideline
compliance improves stroke outcome: a
preliminary study in 4 districts in the Italian region
of Lombardia. Stroke, 3(5): 1341-1347, 2002.
S. Miksch: The Delt/A Project. 2007. Retrieved at
7/4/2011 from the following web site:
http://ieg.ifs.tuwien.ac.at/projects/delta/.
K. Miller and W. MacCaull: Verification of
careflow management systems with timed BDICTL
logic. In: Proc. Business Process Management
Workshops, LNBIP 43, pp. 623-634, 2009.
N. Mulyar, W.M.P. van der Aalst, and M. Peleg: A
pattern-based analysis of clinical computer-
interpretable guideline modeling languages. J Am
Med Inform Assoc, 14(6): 781-787, 2007.
N. Mulyar, M. Pesic, W.M.P. van der Aalst, and M.
Peleg: Declarative and procedural approaches for
modelling clinical guidelines: addressing flexibility
issues. In: Proc. Business Process Management
Workshops, LNCS 4928, pp. 335-346, 2007.
C.P. Neumann and R. Lenz: alpha-Flow: A
document-based approach to inter-institutional
process support in healthcare. In: Proc. Business
Process Management Workshops, LNBIP 43, pp.
569-580, 2009.
C.P. Neumann and R. Lenz: A light-weight system
extension supporting document-based processes in
healthcare. In: Proc. Business Process Manage-
ment Workshops, LNBIP 43, 557-568, 2009.
L.J. Osterweil, L.A. Clarke, and G.S. Avrunin: An
integrated collection of tools for continuously
improving the processes by which health care is
delivered: a tool report. In: Proc. Business Process
Management Workshops, LNBIP 43, pp. 647-653,
2009.
J.M. Overhage, W.M. Tierney, X.H. Zhou, and C.J.
McDonald: A randomized trial of corollary orders
to prevent errors of Omission. J Am Med Inform
Assoc, 4(5): 364-375, 1997.
M. Peleg: Guideline and workflow models. In:
Clinical decision support - the road ahead. R.A.
Greenes, Elsevier/Academic Press, 2006.
M. Peleg, S. Keren, and Y. Denekamp: Mapping
computerized clinical guidelines to electronic
medical records: knowledge-data ontological
mapper (KDOM). J Biomed Inform, 41(1): 180-
201, 2008.
M. Peleg, P. Soffer, and J. Ghattas: Mining process
execution and outcomes. In: Proc. Business
Process Management Workshops, LNCS 4928, pp.
395-400, 2007.
M. Peleg and S.W. Tu: Decision support, know-
ledge representation and management in medicine.
Yearbook of Med Inform, pp. 72-80, 2006.
M. Peleg and S.W. Tu: Design patterns for clinical
guidelines. Artif Intel Medicine 47(1), 1-24, 2009.
M. Peleg, S.W. Tu, J. Bury, P. Ciccarese, J. Fox,
R.A. Greenes, R. Hall, P.D. Johnson, N. Jones, A.
Kumar, S. Miksch, S. Quaglini, A. Seyfang, E.H.
Shortliffe, and M. Stefanelli: Comparing computer-
interpretable guideline models: a case-study
approach. J Am Med Inform Assoc 10(1), 52-68,
2003.
M. Peleg, D. Wang, A. Fodor, S. Keren, and E.
Karnieli: Lessons learned from adapting a generic
narrative diabetic-foot guideline to an institutional
decision-support system. Stud Health Technol
Inform,139: 243-52, 2008.
M. Pesic, H. Schonenberg, N. Sidorova, and
W.M.P. van der Aalst: Constraint-based workflow
models: change made easy. Proc. CoopIS'07, LNCS
4803, pp. 77-94, 2007.
S. Quaglini: Information and communication
technology for process management in healthcare:
a contribution to change the culture of blame. In:
Proc. Business Process Management Workshops,
LNBIP 17, pp. 308-311, 2008.
S. Quaglini, M. Stefanelli, A. Cavallini, G. Micieli,
C. Fassino, and C. Mossa: Guideline-based
careflow systems. Artif Intell Med, 20(1), 5-22,
2000.
S. Quaglini, M. Stefanelli, G. Lanzola, V.
Caporusso, and S. Panzarasa: Flexible guideline-
based patient careflow systems. Artificial
Intelligence in Medicine 22(1), 65-80, 2001.
M. Reichert and P. Dadam: ADEPTflex:
supporting dynamic changes of workflows without
losing control. Journal of Intelligent Information
Systems, 10(2), 93-129, 1998.
M. Reichert, P. Dadam, S. Rinderle-Ma, A. Lanz,
R. Pryss, M. Predeschly, J. Kolb, L.T. Ly, M.
Jurisch, U. Kreher, and K. Goeser: Enabling Poka-
Yoke workflows with the AristaFlow BPM Suite.
Proc. BPM'09 Demonstration Track, CEUR WS
Proc. Vol. 489, 2009.
M. Reichert, S. Rinderle-Ma, and P. Dadam:
Flexibility in process-aware information systems.
LNCS Transactions on Petri Nets and Other
Models of Concurrency (ToPNoC), Special Issue
on Concurrency in Process-aware Information
Systems, LNCS 5460, 115-135, 2009.
H.A. Reijers, N. Russell, S. van der Geer, and G.
Krekels: Workflow for healthcare: a methodology
for realizing flexible medical treatment processes.
In: Proc. Business Process Management
Workshops, LNBIP 43, 593-604, 2009.
S. Rinderle, M. Reichert, and P. Dadam:
Correctness criteria for dynamic changes in
workflow systems - a survey. Data & Knowledge
Engineering, 50(1): 9-34, 2004.
A. Rozinat and W.M.P. van der Aalst:
Conformance checking of processes based on
monitoring real behavior. Information Systems,
33(1): 64-95, 2008.
N. Russell, W.M.P. van der Aalst, and A.H.M. ter
Hofstede: Exception handling patterns in process-
aware information systems. In: Proc. CAiSE'06, pp.
288–302, 2006.
S. Sadiq, W. Sadiq, and M.E. Orlowska: A
framework for constraint specification and
validation in flexible workflows. Information
Systems, 30(5): 349-78, 2005.
S. Sadiq, M.E. Orlowska, W. Sadiq, and K. Schulz:
When workflows will not deliver: the case of
contradicting work practice. In: Proc. BIS'05, pp.
69-84, 2005.
C. Sauer: Why information systems fail. Oxford-
shire, UK, Alfred Waller Ltd, 1993.
R. Serban, A. ten Teije, F. van Harmelen, M.
Marcos, and C. Polo-Conde: Extraction and use of
linguistic patterns for modelling medical
guidelines. Artif Intell Med, 39(2): 137-49, 2007.
Y. Shahar, S. Miksch, and P. Johnson: An
Intention-based language for representing clinical
guidelines. In: Proc AMIA Symp, pp. 592-596,
1996.
Y. Shahar, S. Miksch, and P. Johnson: The Asgaard
project: a task-specific framework for the
application and critiquing of time-oriented clinical
guidelines. Artificial Intelligence in Medicine,
14(1-2): 29-51, 1998.
E. Shalom, Y. Shahar, M. Taieb-Maimon, G. Bar,
A. Yarkoni, O. Young, S.B. Martins, L. Vaszar,
M.K. Goldstein, Y. Liel, A. Leibowitz, T. Marom,
and E. Lunenfeld: A quantitative assessment of a
methodology for collaborative specification and
evaluation of clinical guidelines. J Biomed Inform.
41(6): 889-903, 2008.
S. Shea, W. DuMouchel, and L. Bahamonde: A
meta-analysis of 16 randomized controlled trials to
evaluate computer-based clinical reminder systems
for preventative care in the ambulatory setting. J
Am Med Inform Assoc, 3(6): 399-409, 1996.
P. Terenziani: A hybrid multi-layered approach to
the integration of workflow and clinical guideline
approaches. In: Proc. Business Process Manage-
ment Workshops, LNBIP 43, pp. 539-544, 2009.
S.W. Tu, J.R. Campbell, and K.A. Musen: SAGE
guideline modeling: motivations and methodology.
Stud Health Technol Inform 101: 167-71, 2004.
B. Weber, M. Reichert, J. Mendling, and H.A.
Reijers: Refactoring large process model
repositories. Computers in Industry, 62(5): 467-86,
2011.
B. Weber, M. Reichert, W. Wild, and S. Rinderle-
Ma: Providing integrated life cycle support in
process-aware information systems. Int'l J of
Cooperative Inf Systems, 18(1): 115-65, 2009.
B. Weber, S. Sadiq, and M. Reichert: Beyond
rigidity - dynamic process lifecycle support: a
survey on dynamic changes in process-aware
information systems. Computer Science - Research
and Development, 23(2): 47-65, 2009.
S. Zugal, J. Pinggera, and B. Weber: Toward
enhanced life-cycle support for declarative
processes. J of Software Maintenance and
Evolution: Research and Practice, 24(3): 285-302,
2012.
