25-08-2017, 09:32 PM
MediNet: Personalizing the Self-Care Process for Patients with
Diabetes and Cardiovascular Disease Using Mobile Telephony
1378468185-ref1.pdf (Size: 506.04 KB / Downloads: 29)
INTRODUCTION
Current telemedicine systems have created a means
whereby health monitoring systems have transferred the
point of care closer to the patient and have allowed the
patient to be better informed and actively involved in the
self-care process [1]. Several applications have been
developed where the remote monitoring of a patient’s vital
measurements has allowed better treatment of patients in
isolated areas [2]. This paper describes a system known as
MediNet that is being developed to personalize the
healthcare process for patients with diabetes and
cardiovascular disease using a cellular phone network.
Personalization can be defined as the modification of the
nature of an activity which a user is engaged in or wants to
be engaged in, to the unique characteristics of the user and
the context within which the user currently resides [5].
Personalization in telemedicine is important since every
patient's needs are different [3]. Personalized healthcare
provides an opportunity to achieve more proactive treatment
where problems can be addressed and hopefully prevented at
the earliest possible moment. Reactive approaches on the
other hand, emphasize mainly diagnosis and treatment. The
proposed system will achieve personalization at two levels
one based on group level characteristics that are common to
several patients having the same disease, and the other based
on characteristics that are specific to the individual patient.
The pilot project focuses on patients with both diabetes
and cardiovascular disease. The UK Prospective Diabetes
Study (UKPDS) revealed that better blood pressure control
in diabetic patients who have high blood pressure reduces
the risk of: death from long-term complications of diabetes
by a third, strokes by more than a third, and serious
deterioration of vision by more than a third. Thus patients
suffering from diabetes and hypertension can reduce their
morbidity and mortality by maintaining tight control of their
blood sugar and blood pressure.
In the following sections we present an overview of the
MediNet system. The components of the system are
introduced in section II and the client’s interface is described
in section III. Section IV illustrates some personalized
recommendations made to the patients. Here the reasoning
engine makes use of both the group level characteristics and
individual level characteristics to generate an appropriate
response. Lastly, sections V, VI and VII present the planned
testing, related work, and future research respectively
INFORMATION SYSTEM COMPONENTS
The MediNet architecture, as shown in Figure 1, is
composed of components that achieve four distinct functions
namely: data collection and storage, data reasoning, alert
notification and response, and reporting. It focuses primarily
on diabetes and cardiovascular disease but will be extended
to other medical diseases in later phases.
Glucose and blood pressure sensors are used to measure
the patient’s current glucose level (GL) and blood pressure
(BP) respectively. Custom built Windows Mobile 6 software
on the patient's cellular phone transfers the measurements
from the sensors to the patient’s cellular phone using a USB
or Bluetooth connection. The data is then transmitted from
the phone to a Web server via GPRS. If connectivity is low
the data is stored for later re-transmission. At the Web
server, the sensor readings are combined with historical
readings as well as the patient’s qualitative data and
transferred to the reasoning engine. The reasoning engine
uses this information to generate a personalized course of
action that is appropriate for the patient at the given moment.
Depending on the severity of the condition, the system may
also notify other agencies.
The alert notification component allows medical officers
or care takers to remotely monitor the patient’s condition.
Depending on the situation, an appropriate response action is
initiated such as a phone call to the patient or an onsite visit.
Lastly, the report function processes and presents the
information in several formats depending on the type of
presentation that is required.
Personalization Parameters
The information presented to a patient can be personalized
along various dimensions such as the patient's profile, the
patient's device (i.e., type of cellular phone being used),
connectivity, the patient's context and location, and the
content and goals of the treatment process [5]. The MediNet
system seeks to personalize the information presented to a
patient based on the patient’s profile, the patient's context
and location, and the content and goals of the medical
treatment process. Personalization is achieved by varying the
nature, form, and structure of the information presented to
patients based on the values of the personalization
parameters
When a new patient is added to the system, "static" parameters such as the type of disease, age, sex, severity of
disease, disabilities, socioeconomic position, and lifestyle
are part of the initial data capture. “Dynamic" parameters
such as location and daily activities are input daily by the
patient interacting with the software on his/her cellular
phone. Using the static and dynamic information available
together with the readings obtained from the sensors, the
reasoning engine executes its personalization algorithms.
This results in a “one-to-one” interaction between the patient
and the system, unlike the "one-size-fits-all" approach of
other systems we have surveyed
Process of Personalization
Personalization takes place on two levels in MediNet:
group level and individual level. Group Level
personalization adapts the information presented to patients
based on similar characteristics of the patients (condition –
relevant issues). This is a "macro" form of personalization
and enables coarse-grained personalization decisions to be
made based on general characteristics of the patient
population.
At the Individual Level of personalization, the information
presented is further refined to accommodate individual
differences among patients. At this level, the personalization
is based on parameters such as the patient's lifestyle,
disabilities, prognosis, location, and daily activities. These
parameters vary for different individuals and thus the
personalization process can capture individual differences in
its reasoning processes.
The personalization of the information that is presented to
the patient involves the employment of a transformable base
object whose structure and composition is altered to suit the
personalization dimensions. XSLT and Cascading Style
Sheets (CSS) provide the functionality for modifying the
base object. Apart from the daily health messages that the
patient receives, proactive messages are occasionally sent to
patients to warn them of any activity that may have an
impact on their blood glucose and blood pressure levels.
These messages are also tailored to the individual
characteristics of each patient.
The parameters involved in Goup Level personalization
include type of disease, sex, age group, and the severity of
the disease. Individual personalization parameters are
lifestyle, disability, socioeconomic position, prognosis,
location, and daily activities. The cultural environment
within which the the patient resides, would also be factored
in the Group Level personalization process so as to create an
atmosphere of reality and belonging in the interactions with
patients.
Personalization Example
We now present an example showing the output of the
personalization process given certain attributes known about
two patients. Table 1 illustrates a sample of the
recommendations that would be presented to patients using
blood pressure measurements. Table 2 shows the attributes
of the two patients, derived from the personalization
parameters we are investigating. Given the values of these
attributes, each patient receives a personalized message.
Based on the data presented in Tables 1 and 2,
personalized messages are generated using a mapping of the
corresponding personalization parameters with the current
data. The personalized messages for Patient A and Patient B
PLANNED TESTING
The effects of the system on patient behaviour and
glycaemic/pressure control will be studied in a controlled
trial with 15 patients which will begin in August 2008. The
testing will comprise of patients who have both diabetes and
high blood pressure. All the patients will test the patient’s
interface using the same cellular phone platform where the
functions will be demonstrated as depicted in Figure 2. The
testing group will comprise of adults, both males and
females. The patients are presently playing an important role
during the development of the patient's interface by
providing feedback on the usability of the interface.
When the system is ready for testing, all the participants
will be asked to fill out a questionnaire that will provide
information to guide the group level and individual level
personalization processes. At the end of the trial, all the
patients will be invited to participate in the review of the
pilot project. The results of the testing phase will be
carefully studied to evaluate the accuracy of the
recommendations made by the MediNet system.
RELATED WORK
Reference [2] developed a mobile diabetes management
system for home-care treatment coupled with real-time
patient-doctor communications using an information systems
architecture that is similar to ours. Reference [6] developed a
diabetes support system using mobile phones; in this system,
rather than using sensors, the user must input information
regarding his/her daily condition. Other systems have been
developed to diagnose persons for diabetes given
information supplied such as their age, blood pressure level,
body mass index and the diabetes pedigree function [8].
However, MediNet does not perform diagnosis. Rather, it is
intended to provide patient management and monitoring
capabilities.
A system which provides the base-level functionality of
MediNet is described in [7]. This system was developed by
researchers at the Biomedical Signal Processing Group of
the University of Oxford. It also uses a LifeScan OneTouch
Ultra Blood Glucose Meter to obtain blood glucose readings
from patients. The communication between the patient's
mobile device and the server is made using GPRS. However,
the system only deals with Type I diabetes and does not
provide the level and depth of personalization that our
MediNet is designed to handle.
Although mobile telemedicine is a rapidly growing
research field, personalization research in telemedicine has
not yet received the level of attention it deserves although it
can make a significant influence to the improvement in
quality of care provided. The added value of the MediNet
system to current systems is that it seeks to incorporate
personalization as an important dimension in management of patients. Personalization would serve to improve the quality
of patient-centered health-care by providing ongoing
relevant feedback to the patient thereby encouraging and
eventually improving the patient’s self-care management of
diabetes and/or cardiovascular disease.
CONCLUSION AND FUTURE RESEARCH
This paper described the MediNet system that is being
developed to provide personalized recommendations to
patients suffering from diabetes and high blood pressure in a
mobile environment. We believe that personalization of the
self-care process can make the process more patient-friendly
and encourage patients to take care of their condition in a
more effective way that will lead to longevity and reduced
risk of complications. Based on the results of the testing in
August 2008, we will determine the usefulness of the
personalization process and introduce additional parameters
to make the personalization more effective. Future research
will involve a study of how cultural context can be taken
into consideration during personalization. Investigations will
also be carried out on extending MediNet to include other
countries of the Caribbean as well on developing a low-cost
implementation for the economically disadvantaged. The
distributive nature of the regional MediNets is likely to
generate an array of research issues involving data
ownership and regulation controls.