Information Technology Reference
In-Depth Information
was also possible to detect the agitation state of a
subject even if the device was not trained on that
specific subject. In addition, it was possible to
define and detect the transition state of a patient.
In order to achieve that, a new confidence measure
on the decision of a support vector machine was
introduced (Sakr et al., 2010).
However this detection was done offline and
not using a portable device. The focus of this
chapter is the device designed and implemented for
portable real-time agitation detection. To the best
of our knowledge, there does not exist any other
portable wireless device capable of autonomous
agitation detection.
ables it to stick to the skin of the patient and acquire
accurate measurements. This RTD is linear and
of class A which gives a temperature coefficient
of 0.00385Ω/Ω/
o
C and a drift per year of 0.2
o
C.
Note that agitation detection has to be fast thus
this sensor was chosen because its response time
is less than 1 second. The GSR sensor is simply
formed of two electrodes that wrap around the
fingers. The heart rate sensor is the polar chest
sensor from Vernier. This sensor is powered by
a 5V DC source and generates a 1.8V pulse with
every heart beat. A strap is placed on the chest of
the subject and the pulse is sent wirelessly to the
receiver placed next to the microcontroller. Since
the heart rate sensor has limited power capabilities
as well as a pulse of 1.8V which is considered to
be low for the microcontroller, signal conditioning
circuits are needed for the heart rate sensor. And
also circuits are needed to compute the resistances
of the RTD and the GSR sensor. Figure 1 shows
a photo of the sensors described in this section
along with the device.
DEVICE AND ALGORITHM DESIGN
In this section the device design along with the
detection algorithm are detailed. The hardware
developed will be discussed in details. The algo-
rithm itself, which was previously developed by
our team and detailed in Sakr et al. (2010), will
be briefly presented.
Signal Conditioning
Device Design
For the heart rate sensor, a simple non-inverting
unipolar amplifier is used with gain G=11. Since
the input is 0V when there is no beat and 1.8V
when there is a beat, the signal is amplified to 0V
when there is no beat and +Vsat=5V when there is a
pulse. This signal is then fed to the microcontroller.
As for both RTD and GSR, a normal Wheatstone
bridge is used. The three resistors used for the
RTD bridge are fixed resistors of 100Ω. These
resistors are regulated such that the output of the
bridge is 0V at 0
o
C. As for the GSR electrodes,
the bridge uses three resistors of 500KΩ. Figure
2 shows a photo of the device prototype used for
testing. Figure 3 shows a top view photo of the
final device developed. Figure 4 shows a detailed
electric diagram of the conditioning circuits and
their connection to the microcontroller.
In this section, the wearable device is discussed.
First the sensors used are presented along with
their corresponding conditioning circuits then
the microcontroller and the decision algorithms
are introduced.
Sensors
As stated previously the three vital signs measured
are the skin temperature, the GSR and the Inter-
Beat Interval (IBI). The choice of these specific
signs is analyzed in the next section. To measure
the skin temperature, the Omega SA1-RTD sensor
is used. It has an accuracy of 0.15
o
C at 0
o
C. This
sensor has surface mount characteristic which en-
Search WWH ::
Custom Search