Biomedical Engineering Reference
In-Depth Information
have any of these models been checked for their applicability to cardiopulmonary
patients nor do specialized HR models exist for these.
Song et al. introduced a set of rules to control the training performance of COPD
patients, Nee et al. used Bayesian networks for an adaptive alert system for patients
with heart issues [8], Schulze et al. used Bayesian networks for the training control of
COPD patients [9]. These machine learning based approaches showed to be well
suited for adapting systems to the individual differences between users, but they have
to be trained first. The rule based system is a more abstract definition that fits for most
of the users, but uses hard coded thresholds which are not individualized. There exists
no model that can be used to define individual reference values for the rule based
thresholds or initial / start values for the machine learning based approaches.
Individualized models have to be able to take all values into account, which may
be relevant for the prediction. Multiple professional maintained Electronic Patient
Records (EPRs) containing data from one patient are typically maintained in several
different health service institutions, storing different data in different formats and
providing a variety of proprietary interfaces. The user controlled Personal Health
Record (PHR) is meant to overcome these limitations by providing standardized
communication interfaces, thus enabling data exchange in the course of medical
treatments among various institutions.
Many PHR products with a wide range of features were introduced in the last ten
years, but lately some of the big companies removed their products from the market
(e.g. Google Health, ICW LifeSensor), because the usage fell short of the expecta-
tions. In [10] the Markle Foundation introduced PHR architectures. The typical archi-
tecture for PHRs is a web-based third-party tethered PHR, where a user enters and
uploads his health related data to a web server, which is maintained by private com-
panies like Microsoft or Google. A survey of the California HealthCare Foundation
[11] revealed that 55 percent of patients with chronic conditions are concerned about
the confidentially of their health data. As privacy concerns inhibit the adoption of
PHRs, it is critical to obtain broader acceptance. Interoperability is a critical prerequi-
site to connect the users PHR with professional EPRs, where most of the user´s health
data is stored. We compared web-based PHR products in [12] and found that only a
minority of the existing PHR systems make use of existing standards that enable inte-
roperability for data exchange.
1.3
Aim and Scope
HR is an important vital parameter and thereby an important indicator of a patients
physical state during rehabilitation training [13]. The knowledge about factors that
have an influence on the exercise physiology might help physicians and autonomous
systems to take this information into account when deciding how much load a patient
can undergo during a training session. Hence it could be used to support creation and
optimization of training schedules and during the current training session itself to
derive the future course.
A difference between the predicted trend of a normal training and a measured heart
rate may give a hint on a potentially abnormal development and thereby help to detect
critical states before they occur. This is especially important in telerehabilitation
settings, where patient's train under unsupervised conditions at home (see [14, [15]).
Search WWH ::
Custom Search