Technical details of the Wearable module
The heart of the wearable module is the Parallax Propeller microcontroller (μC).
The controller is designed with 8 cores running at 80MHz.
The multi-core μC can run parallel tasks simultaneously.
Our application requires parallel data acquisition,
signal processing,
wireless data transmission and display.
The module has a 4 channel 10-bit ADC,
sampled at 1.2 kHz (300 Hz per channel),
and a 2-channel DAC.
A custom designed for the project version of the open hardware architecture of Propeller Board of Education (BOE) has embedded the XBee module in the design.
Since the XBee communication is bi-directional the remote client can perform simple remote tasks,
such as:
- resetting the wearable module;
- pausing acquisition;
- requesting and scheduling NIBP measurements;
- adjusting algorithms’ coefficients;
- and enabling or disabling the wearable display.
To simplify user interface,
the patient need only to operate only an on/off switch.
Once the system is on,
acquisition and transmission of data is automatic.
The wearable system is powered by a 9V Li-Ion rechargeable battery (500 mAh).
The system consumes approximately 60 mA with display unit turned off and a single NIBP measurement per hour.
Consumption is affected by the frequency of NIBP measurements and display unit utilization.
The consumption climbs to 140 mA with continuous use of the display.
The unit under low power consumption setup is capable of continuous measurements and transmission for 8 hours.
The ECG module is capable of acquiring a single Einthoven lead (I,
II or III),
depending on electrodes placement.
The ECG signal is acquired from two chest electrodes connected with a shielded cable to the module [10]-[12].
The ECG signal is acquired,
amplified and filtered with the use of passive and 2nd-order active filter in the frequency range of 0.5 to 150 Hz.
The filtered signal is sampled at 300 Hz,
satisfying Nyquist theorem and providing an ECG signal of sufficient quality for diagnostic purposes.
The patient’s Heart Rate (HR) is calculated through an appropriate derivative R-wave detection algorithm.
The R-wave is detected when signal’s amplitude exceeds a threshold Θ (frequently adapted by software,
as a percentage of the recorded peak amplitude) and when there is a fast change in signal’s slope (signal’s derivatives).
The arterial pulse Oximetry signal is acquired with a Nellcor finger sensor.
A DAC interfaces to a LED-driving circuit,
adapting Red and Infrared (IR) LEDs current for sufficient absorption signal.
The finger’s photodiode sensor signal is amplified (employing a differential current sensing trans-impedance configuration),
filtered and sampled at 300 Hz,
utilizing two ADC channels,
one for each wavelength.
SpO2 is calculated from the averaged natural logarithm ratio of RED and IR amplitudes (Peak and valley method).
The NIBP measurement is performed according to an oscillometric technique.
A wrist-cuff is inflated to a clinician predefined pressure.
During deflation the cuff pressure and pressure oscillations are monitored.
Initiation of oscillations marks systolic blood pressure (BP),
while termination marks diastolic blood pressure.
During NIBP measurement SpO2 detection is suspended.
Both NIBP and SpO2 sensors are located on the same hand resulting in false SpO2 alarms,
during NIBP blood flow occlusion [13],
[14].
Laptop server and remote client
The laptop server hosts the USB Xbee stick,
which provides the wireless communication to the wearable module and a web application developed with Microsoft.NET technology.
The Web application receives quasi-real time data,
transmitted by the wearable module and presents data’s waveforms (single lead ECG and Plethysmography),
numerical values (HR,
SpO2 and BP) and as physiological parameters numerical trends.
The basic signal processing,
such as signal filtering,
HR and SpO2 calculation is performed on the wearable device.
Access to remote viewing is provided by user authentication.
Authenticated users are allowed to:
- change alarm limits for the monitored parameters;
- apply remote operations on the wearable module,
such as resetting the device,
enabling and disabling wearable display;
- initiating a NIBP measurement;
- changing NIBP measuring intervals.
The server allows for the future employment of more demanding algorithms,
still under development.
System’s preliminary laboratory evaluation
XBee communication was tested with a walk test outside the laboratory,
at small distances and the signals were transmitted through brick walls for over 30 m.
The functionality of the wearable module has been tested mainly with the use of patient’s simulators.
The ECG acquisition,
as well as,
HR detection algorithm were tested with a BIO-TEK Instruments Inc patient simulator (Lionheart multi-parameter simulator).
The developed HR detection algorithm was found to accurately detecting HR at 30,
60 and 120 bpm.
HR was calculated by an averaged R to R interval of five pulses.
Similarly,
the SpO2 accuracy was tested with a Fluke Biomedical finger SpO2 simulator (Metron).
Oxygen saturation measurement was tested in the range of 80 to 99%,
with an accuracy of +/- 2% at a HR of 60 bpm and a motion free setup.
Due to lack of a NIBP simulator,
the NIBP measuring module was tested on the authors.
The NIBP measurement was performed successively by the wearable device and a blood-pressure measuring equipment based on Korotkoff’s sounds method.
Although there was no significant difference (+/- 5 mm Hg) between the two methods,
there are several limitations. The detection of Korotkoff’s sounds is subjective.
The measurements were not tested against hypertension and hypotension clinical scenarios.