|
We propose adopting a systematic and
integrated approach to the design and provision of
information to clinical oncologists and breast cancer
patients in a routine clinical environment. The basis of the
proposal is to pool the required multidisciplinary expertise
to design two specific information support, software tools
that link to the prognostic information available about the
individual patient, and present this in a user-centered,
flexible way, so as to maximise the utility to the
oncologist and to the patient. This requires skills in
intelligent systems, bioinformatics and interface design.
The eventual systems follow the main requirements of the
Medical Devices Directives, in being design assured by use
of best practice in statistical, neural and expert system
inference, together with risk assessment, to verify the
inferences made against clinical knowledge and also to
assure the integration of the software system as a whole,
and finally an evidence-based evaluation of effectiveness to
the two types of users. This approach complies with a solid
framework for the development and evaluation of complex
interventions to improve medical care (http://www.mrc.ac.uk/complex_packages.html).
Such an approach, unlike traditional
decision-making techniques including multi-criterion, will
provide breast cancer clinicians and patients with a
high-assurance, decision support that is adaptive to their
decision practices yet allowing for evolutions of decision
models, decision resources (data) and other users concerns.
This novel approach will provide important insight into the
development of an integrated decision support infrastructure
for high-assurance decision activities.
Required Research
The evidence-based approach to the
delivery of medical care has gained wide recognition within
the healthcare community, advocating that decision-making
should use current knowledge and clinical evidence from
systematic research (Sackett et al 1996). In breast cancer
care, there are currently few staging methods in widespread
use by clinicians, namely the Nottingham and Manchester
staging systems. Patients are provided with key information
mainly during one pre-set appointment with the clinician and
perhaps the breast care nurse. This can lead to patients
being given information that they are not able to
rationalise, or is pitched at an inappropriate level of
detail, perhaps also missing out on the opportunity to ask
for further clarification as doubts and queries set in after
the event. In both the case of the patient, and of the
clinician, there is a need not just for accurate and
relevant information, but also to make this accessible at
the time when the user requests it, in a form that the user
will relate to, and with the opportunity to follow-up levels
of detail under control of the user.
Enabling such
decision-making processes will require addressing a number
of issues including:
-
How to integrate support models for clinicians and for
patients within a common reasoning framework.
-
How to identify and fetch evidence that is relevant to a
specific clinician?s and/or patient?s decision point, and to
validate the advice given by recourse to explicit reasoning
models.
-
How evidence and decision traceability is best presented
to each group of users.
To address these questions, we will study and model the
information needs of clinicians and patients, developing
techniques tailored to the level of complexity of the users,
but providing also the capability to trace the advise given
in progressively greater levels of detail. While the goals
we have set ourselves are ambitious, we believe that they
are realisable in the proposed timescale of 3 years, as
follows:
Given the multi-disciplinary nature of the specification,
design, implementation and evaluation aspects of this health
informatics system, the project will actively involve
clinicians, statisticians, computer scientists and public
health researchers, who will participate in the modelling,
specification development and evaluation of the proposed
work. In short, it will require active participation by the
complete patient care team, in dialogue with psychologist,
sociologist, statisticians and computing specialists who
will implement the informatics system. The expertise to do
this is currently in the consortium.
-
The starting point for the provision of patient specific
information are prognostic groups allocations by statistical
survival models, together with rule-based reasoning captured
from expert clinicians and other members of the care team.
This will progress to include an interactive user interface
supported by a reasoning model, which sophistication at
first will be kept as simple as possible to facilitate
verification and validation of informatics system as a
whole, and building user acceptance and trust in the system.
It will then be augmented with a visualisation tool with
self-organising data maps and data segmentation facility.
-
The survival and segmentation models may be used to define
metrics for data access. Principled statistical and pattern
recognition methodologies to model censored data and for
visualisation and segmentation are already in the domain of
expertise of the core team of researchers, as are the
aspects of rule-based reasoning and multimedia engineering.
|
Project Title
Towards a Disciplined Approach to
Integrating Decision-Support Systems for Breast Cancer Care
Activities
Sponsors
Polaris House,
North Star Avenue,
Swindon. SN2 1ET
Partners
Wilmslow Road,
Manchester, M20 4BX.
Clatterbridge Road
Bebington
Wirral. CH63 4JY
School of Computing and Mathematical Sciences
Liverpool John Moores University
James Parsons Building
Byrom Street
Liverpool
L3 3AF.
Tel (+44) 1524 593801
|
