Dr. Wendler
is a Senior Scientist in Philips Research Laboratories, Hamburg,
Germany.
W
e live in a world of rapid change. In the healthcare system--and in
radiology as an integral part of it--we are facing a challenging
situation of shrinking budgets, increasing cost pressure, and
growing demands to increase the efficiency and quality of services.
Reacting to these challenges, healthcare enterprises increasingly
rely on Information Technology (IT) solutions. In digital
radiology, IT environments integrate various application systems:
hospital information systems (HIS), radiology information systems
(RIS), picture archiving and communication systems (PACS), and
imaging modalities. These systems provide integration on the
information level, and are based on working standards that support
data exchange (DICOM [Digital Imaging and Communication in
Medicine] and HL-7 [Health Level 7]). Access to information is an
important aspect for radiology, but not the only one. The
transition from film-based to digital imaging systems is more than
a change of technology for storage, distribution, and display of
data. Simultaneously, we see that the
processes
in radiology and the
organization
of work change as well. Even more, under new socioeconomic boundary
conditions, healthcare institutions are forced to optimize or
completely redesign departmental processes to meet their essential
medical and commercial targets. Increasingly, business process
re-engineering (BPR) and process automation (Workflow Management)
are seen as key factors for the successful operation of digital
imaging departments and hospitals. This is in line with
observations from other industries (e.g., banking or insurance),
which are related to medicine in the sense that professionalism and
success are based on a strict approach to customer orientation and
cost effectiveness.
Managing processes: Where are we now?
Current IT solutions for radiology are rarely designed to
explicitly and dynamically support changing processes and flexible
organizations. Reports on efficiency improvement and cost reduction
after IT installations do not sound overly enthusiastic. Usually,
workflow aspects are dealt with in time-consuming and expensive
projects, rather than in products that would support this important
aspect of system integration. We lack effective concepts to deal
with business processes on the departmental level. Functionally,
workflow aspects are spread over the entire information system
environment, with clear drawbacks, such as: 1) preventing an
overall and integrated view of processes and organizations;
2) creating overlap of the existing workflow functionality
implemented in the various subsystems, especially in multi-vendor
environments; and 3) preventing an adaptation to the specific and
changing needs of the institution (process re-engineering and
optimization).
What we have achieved so far is integration on the data level,
and the standards that support data exchange have improved
significantly over the past few years. DICOM, however, the
essential "data-driven" standard for radiology, offers very little
support for workflow. We have seen attempts to extend DICOM in the
workflow direction; however, these extensions do not appear general
enough to meet requirements of future workflow-driven systems.
What kind of systems do we envision?
There are demands to move forward and take the next logical step
from pure data management to an environment with integrated data
and workflow services. What we want are information systems that
are explicitly designed to: 1) contribute to efficiency,
cost-effectiveness, and a high standard of medical
services; 2) flexibly support the key processes and organization in
changing medical institutions; and 3) support the radical
re-engineering, permanent optimization, and automation of processes
to meet economic and medical needs.
Workflow Management Services with model-based enactment of
workflow have to be integral parts of medical IT solutions.
Workflow architectures and standards are needed to support
this.
What is workflow?
Unfortunately, the term workflow is frequently misused to
describe certain aspects of data flow. We suggest using the
standard workflow vocabulary as defined by the Workflow Management
Coalition,
1,2
the most important standardization body in this area:
workflow is the automation of a business process, in whole or
part, during which documents, information, or tasks are passed from
one participant to another for action, according to a set of
procedural rules.
Workflow deals with processes, and the flow of data is just one
aspect of it. Other aspects include: Modelling the sequence of
activities and subactivities (process models); modelling all
aspects of human resources (organizational models); management of
operational units (equipment, devices, tools), locations (rooms,
workplaces), and materials (consumables); organization and tracking
of patient flow; and the adherence to all kind of medical,
political, economic, legal, and time constraints.
Figure 1 depicts an extremely simplified typical
workflow-centric view of radiology that emphasizes the process
aspect.
3
The left and right sides of the diagram show application system
components implemented as functional modules of RIS, PACS, and
imaging modalities. As shown, these components are needed in
well-defined activities of the diagnostic process. In the middle
the process itself is illustrated, represented by explicit process
models that are enacted at run time using a workflow management
system (WfMS). In figure 1, the conceptual independence of an
autonomous workflow management service and the application systems
becomes obvious.
At run time of an IT environment, previously defined process
models are selected, initiated, and enacted. In radiology, a
typical trigger event to start a particular process is the
registration of a request for a specific kind of radiological
study. After process start, the sequence of events follows the
process model accordingly. The activities that represent each step
are either fully automated (e.g., pre-fetching images, which is
performed by a PACS component) or done manually by workflow
participants. In the latter case, work items are passed to workflow
participants in terms of worklists, which are presented to users in
the appropriate look-and-feel of their dedicated application
system. Typical examples are worklists like "Cases to be reported"
in a softcopy diagnosis PACS workstation, or "Scheduled patients to
be registered" in a RIS.
A workflow service can be built using workflow management
technology and commercial off-the-shelf WfMS. In general, these
systems allow users to define, create, and manage the execution of
workflow through the use of software, running on one or more
workflow engines, which are able to interpret process definitions,
interact with workflow participants, and, where required, invoke
the use of IT tools and applications. Some basic terminology issues
for WfMS are presented in figure 2.
In particular, workflow management systems provide: 1)
build-time tools to model processes in terms of activities and
state-transitions, and model the organization in terms of
organizational units and workflow participants; and 2) run-time
tools (workflow engines) to match workflow participants and
activities, put processes into action, and provide worklists
according to the state of processes.
Until now, workflow management systems have proven their
usefulness outside the healthcare domain. There are more than 100
commercially available WfMS on the market, with very different
implementation strategies and modelling paradigms.
Application-independent and process-oriented systems look promising
as candidates for use in radiology.
Workflow-enabled application systems
Following the workflow approach, future application systems need
to fit into workflow architectures. These
"workflow-enabled"applications need--in addition to data-oriented
(DICOM) interfaces--connections to the department's or the
hospital's workflow enactment service. A set of standard interfaces
has been defined by the Workflow Management Coalition.
2
These interfaces have to play an essential role in workflow
architectures, although up to now they have only been implemented
by a few WfMS manufacturers and still have to mature. The
definition and consolidation of workflow architectures is certainly
one of the next steps for the introduction of workflow management
in radiology.
Conclusion
Fully digital radiology departments potentially work more
efficiently, at a lower cost, and offer new options to improve the
services. Introducing information technology as such, however, will
not automatically improve processes. Current IT solutions in
radiology are predominantly data-focused. For the challenges of the
future we need process integration in addition to data integration.
The design of information systems needs to support re-engineering
and process automation. As a starting point, it will be essential
to think "process-oriented" and understand healthcare processes,
which requires a deep knowledge of current processes and an
awareness of changes and future trends that will shape and automate
future processes. Workflow management appears to be a promising
approach to support processes in the healthcare domain, and a
valuable contribution to increase efficiency and reduce costs. It
appears attractive for the automation of the majority of
structured, routine processes in radiology. For processes of ad-hoc
nature (e.g., in emergency situations), it must be complemented by
groupware concepts.
WfM can be implemented by Workflow Management Systems, which
should be integrated as separate, autonomous system components.
They should be independent of the capabilities and limitations of
the attached application systems (RIS, PACS, and imaging
modalities), and should provide scalability to the enterprise level
(as radiology is a sub-process within a healthcare enterprise).
Autonomous workflow services will facilitate the implementation of
workflow management in multivendor environments, and will enable
the use of commercial off-the-shelf Workflow Management Systems
with powerful build-time and run-time tools. This concept calls for
common architectures and standardized interfaces between the WfMS
and the various "workflow-enabled" application systems. It is
recommended to adopt and apply the existing standards defined by
the Workflow Management Coalition, in addition to the data-oriented
standards we already have.
AR
References
1. Lawrence P:
Workflow Handbook 1997, Chichester, England, John Wiley & Sons
Ltd., 1997.
2. Workflow Management Coalition:
http://www.aiim.oriz/wfmc/.
3. Wendler T, Meetz K, Schmidt J:
Workflow management systems in radiology. Proc SPIE Med Imaging
1998 3339:216-225, 1998.