Proceedings of the 9th IEEE International

Conference on Nano/Micro Engineered and Molecular Systems

April 13-16, 2014, Hawaii, USA

Integration of Micro Sensors with Mobile Devices for

Monitoring Vital Signs of Sleep Apnea Patients
*
Rong Zhu , Member, IEEE, Zhe Cao, Ruiyi Que
State Key Laboratory of Precision Measurement Technology and Instruments, Department of Precision Instruments, Tsinghua
University, China
* rong_zhu@263.net

Abstract- Body Sensor Networks (BSNs) integrating micro

sensor nodes with wireless communication are becoming recording. The lab-attended PSG examination is complex,
flourishing in medical applications. This paper demonstrates a time-consuming, costly and relatively invasive [2]. While
BSN-based portable monitor integrating micro sensors with using BSN technology in diagnosing and monitoring of OSAS
mobile devices for monitoring and diagnosing obstructive sleep is an effective way to access healthcare of the whole people at
apnea syndrome (OSAS) at home. The system uses a micro hot- a low cost and with high efficiency, and thus reduces the
film flow sensor to detect respiratory flow, uses tri-axis micro
accelerometer to detect body posture and motion intensity, and threats to the human life.
uses a micro photoelectric sensor to detect oxygen saturation of In this paper, we propose to integrate micro sensors with
blood. The real-time vital signs are detected and transmitted mobile communication devices to develop a BSN-based
wirelessly to a mobile device. The transmitted data are processed portable monitor for OSAS. Three kinds of micro sensors are
and analyzed in the mobile device and can be further transmitted
utilized, including hot-film flow sensor, tri-axis accelerometer,
to a remote medical center and physicians through mobile cellular
networks or Internet for comprehensively analyzing, assessing, and oximeter. The vital signs detected via the sensors are
and diagnosing to the diseases. transmitted wirelessly through Bluetooth to a mobile device,
such as cellphone and PDA, and can be further transmitted to
I. INTRODUCTION
the remote medical center via Internet. The configuration of the
The rapid growth in micro/nano sensors, low-power system is shown in Figure 1.
integrated circuits, and mobile communication has enabled a
II. OPERATION OF SYSTEM
new generation of wireless body sensor network, now used for
purposes such as monitoring health and diagnosing diseases [1]. A. Construction of System
The BSNs field is an interdisciplinary area which could allow The BSN-based system comprises three layers of network as
inexpensive and continuous health monitoring. A number of shown in Fig. 1 and Fig. 2 including the layer of body sensor
MEMS/NEMS sensors can be integrated into a wearable BSN, node, the layer of portable terminal (i.e. mobile communication
which can be used for computer-assisted rehabilitation or early devices) and the layer of remote monitoring center [3]. Sensor
detection of medical conditions. node layer is composed of a respiration-posture sensor node
Obstructive sleep apnea syndrome (OSAS) is a sleep and an oxygen sensor node; wherein using a homemade micro
disorder characterized by abnormal pauses in breathing or hot-film to monitor respiratory airflow, using a 3-axis MEMS-
instances of abnormally low breathing during sleep that based accelerometer to monitor body posture, using a pulse
seriously threatens human life, especial for elderly people. oximeter to monitor blood oxygen saturation; each sensor node
According to medical survey, about 20~40% of elderly people transmits data to a mobile device through Bluetooth. In the
suffer OSAS, and more than half of them remain undiagnosed. mobile communication layer, mobile device acquires data from
Diagnosis of OSA is often based on a combination of patient the sensor nodes in real time, executes the extraction of vital
history and tests. These tests range in decreasing order of cost, signs, and performs a primary diagnosing and early warning
complexity and tethering of the patient (number and type of for severe diseases using auxiliary software, and also can
channels of data recorded), from lab-attended full transmit the data to the remote monitoring center via Internet
polysomnography (PSG) down to single-channel home for executing professional diagnosis by physicians.

Fig. 1. Configuration of BSN-based OSAS monitor system.

978-1-4799-4726-3/14/$31.00 © 2014 IEEE 462

 Respiration-Posture Sensor Node
Hot-film Flow Sensor
Signal Sample and
Bluetooth Cellphone or PDA
Data Process
Transmitter
Tri-axial Accelerometer Data Process and Analysis

Bluetooth Receiver

Oximetry Sensor Node Result Display Data Transfer to Remote Center

Signal Sample and
Bluetooth Transmitter
Oximeter Data Process

Fig. 2. Composition of the monitor system. Filtering

A commercial tri-axis MEMS accelerometer (Analog

B. Respiration-Posture Sensor Node
Devices, ADXL345) is selected as the posture sensor.
Flow *alueCL/min3
There are two sensor nodes in the monitoring system. We
developed the respiration-posture sensor node by ourselves and C. Software for Extracting Vital Signs
used a commercial oximeter (CONTEC Medical Systems, We developed auxiliary
Timing software operated in the mobile
CMS50FW) as the oximetry sensor node. The circuit and the device, which include algorithms of extracting vital signs and
configuration of the respiration-posture sensor node shown in diagnosing/evaluating the disease of OSAS. The derivation
Fig. 3 are tailor-designed aiming at low power consumption processes of respiratory parameters, such as minute
Peak–Detegtion ventilation
Integration
and miniaturization. The used micro hot-film flow sensor is (MV), peak inspiratory flow (PIF), respiratory rate (RR), and
Proge››
fabricated using our previously reported technology [4] tidal volume (TV), are shown in Fig. 4.
PIF RR M*
incorporated film sputtering with printed circuit technique, Sampling o£ Flow
specifically a composite of films 3mm×300μm with Cr/Ni/Pt Data
Sen›or
AGqui›ition Divi›ion
is constructed by sputtering Cr (30nm), Ni (100nm), and Pt
(50nm) layers in sequence on and between the preprinted Au Bluetooth Hot-
film Airflow
electrodes connected with lead wires and welding spots, finally Hot-film sensor
T*
a thin parylene layer is deposited on the wafer as encapsulation. 7mm
T
The flow sensor is operated in constant temperature difference h
e Re›piratory Parameter›
(CTD) mode [5] and uses a self-tuning temperature
8.4mm

r
m
compensation technology [6] to overcome the temperature drift. a
l
The flow range of the hot-film sensor is 0-100L/min, the
threshold Flow-acceleration sensor circuit
reaches 0.01L/min, and the accuracy reaches 1%FS. Flow *alue Di›play
f

The benefits of using the micro hot-film flow sensor lie in: low i Wave£orm Parameter›
Battery Di›play l
Di›play Di›play
cost, easy fabrication and packaging, high sensitivity, m
,

integrating with processing circuits; capability of detecting

comprehensive respiratory parameters (minute ventilation, tidal C
r
volume, etc.), being extremely sensitive to weak flow so that /
N
feasible to detect infants’ respiration. i
/
P
t
/
P
a
r
y
l
e
n
e

0
.
3
m
m
×
3
m
m
×
2
0
0
n
m
e
l
e
c
t
r
o
d
e

463
 e
r seconds or decreases by 30 collapse of the airway.
w
i
i
n
percent if there is an Lateral positions (sleeping on
r g associated decrease in the a side) as opposed to supine
e
oxygen saturation or an positions (sleeping on the
arousal from sleep [7]. The back) are also recommended
p
s a diagnosis of sleep apnea is for sleep apnea, because the
o d based on the quantity of gravitational component is
l polyimide substrate
d apneic and hypopneic events smaller in the lateral position.
per hour of sleep called Therefore, the acquisition of
Apnea Hypopnea Index posture information besides
(AHI), associated to a formal information of breath and
threshold, above which a blood oxygen helps to know
patient is considered as comprehensive sleep
ReGord suffering from sleep apnea, Data symptoms
Regording in diagnosing
and the severity of their sleep OSA.
apnea can then be quantified.
Data Pagking The testing experiment is
and tran›mitting
Tran›mit Mild OSA ranges from 5 to conducted as shown in Fig. 6.
14.9 events per hour of sleep, The tester wears the
Fig. 3. Prototype Fig. 4. Flow diagram for
extracting respiratory moderate OSA falls in the respiration-posture node on
of respiration-
posture sensor parameters. range of 15–29.9 events per the upper body and collects
node. hour of sleep, and severe the respiratory airflow using
Regord
The signs of body posture ReGord OSA would be a patient a nasal cannula. The oximeter
and oxygen saturation of
blood are extracted Data
fromPagking
the and tran›mitting
readouts of tri-axis Tran›mit

accelerometer (ax, ay, az) and

the oximeter respectively as Fig. 5 Flow diagram for extracting
shown in Figure 5. The body posture and blood oxygen.
directions of ax, ay, az refer to
the right, feet, and back of the III. EXPERIMENT
body. The posture signs are S AND having over 30 events per clip is worn on the tester’s
simply categorized into DISCUSSIONS hour of sleep. Severe OSA fingertip. All sensor data are
sit/stand, lie on back, on face, can lead to death by transmitted wirelessly
on right, and on left, generating diurnal through Bluetooth
determined by simply hypertension and severe communication to a mobile
synthesizing the sign and cardiovascular health phone nearby in which the
orientation of the component implications when apnea data are processed and
in ax, ay, az that with the remains undiagnosed and is analyzed for diagnosing
maximum absolute not properly treated [8]. OSA.
acceleration. For example, if Untreated obstructive sleep
ay possesses the maximum apnea may lead to high blood Fig. 6 Schematic view of the
absolute value, when ay is pressure, cardiovascular monitoring for OSA.
positive and oriented to the diseases such as heart failure,
feet, and the posture sign is stroke and heart arrhythmias. Fig. 7 shows the outputs of
determined to be “sit/stand”. Therefore diagnosis of this three sensors in the
By using this simplified disorder at an early time is experiment. During the
method, the computation load critical. Diagnostic tests experimental test, the tester
for the processor is include oximetry and AHI purposely held his breath two
minimized. monitoring in a sleep clinic. times to simulate apneas. At
Medical statistics have around 7th minute, the tester
Sampling o£ 3–a×i› proved that many people swung his body for a while,
Sampling o£
aggelerometer (a×,ay,a›) 0×imeter benefit from sleeping at a 30- which caused a rise of
Data degree elevation of the upper respiratory flow rate and the
Agqui›ition
body or higher, as it helps output fluctuations of
Filtering prevent the gravitational
OSA is the most either an apnea, characterized
common category of by complete cessation of
sleep-disordered
airflow for at least 10 seconds,
Aggeleration(g) breathing. An "event" can be
464
S*M=(ax2+ay2+az2)
 Proge››
N
>thre›hold?
T
Motion
or a hypopnea in whichRe›taccelerometer. The signs of Apnea or
0.
airflow decreases by 50 respiratory flow parameters, 1 hypopnea
percent for Motion
10 oxygen saturation of blood

Quantity
Ma×imum Ab›olute
0
Inten›ity Aggeleration 0 2 4
and pulse rate, as well as 8 10 12
0
11 12 13 14
10
body posture are extracted
Body Po›ture
15 16 17
using the algorithms shown Tidal Volume Duration Of
0.02 2
in Fig. 4 and Fig. 5. The apnea or
TV (L)

Quantity
0.01 0
results are illustrated in Fig. hypopnea
0
Blood 0×ygen 0 (s)
Po›ture Data 8~10.
Wave£orm
And Pul›e Di›play 0 2 4 0
Di›playDi›play Di›play Distributi
8 10 12

Acceleration (g)
Respiratory Flow Acceleration 1
Flow (L/min)

0.2 1 on of
Apnea
ax 0
0.1 0 ay az 0 respirator
0 0 y rate
0 2 4 6 -1
0 2 4 6 8 0
8 10 12 0 10 2 12
4 6
Peak
100 SpO2 0.2 Posture
PIF (L/min)

(g) Code
Insp 8 10 12 14
SpO2 (%)

Sit Lie 0
98 irato
16

SD of SVMPosture
96 ry ³it
0.2 Flo R
0 2 4 6 w 2 e
0 4 6 8 10 12
0.1 s
Acceleration (g)

8 10 12
1 Motion Intensity p
0 Acceleration 0 i
ax ay az Motion r
-1 a
Motion t
0 2 4 6 0 2 4 6 8 10 12 o
8 10 12 Time (min)
r
Time (min) y

Fig. 7 Outputs of flow sensor, R

oximeter, and accelerometer. a
Fig. 10 Extracted body posture and t
R e
e motion intensity.
Flow (L/min)

s
p It is known that the most (
i b
r important index for p
a diagnosis of OSA is AHI m
t )
o associated with the decrease
r in the oxygen saturation. 0.1 Descent of
y 5 SpO2
Therefore we make 0
Quantity

0 2
F statistical analyses on apnea 8 10
l 4 12
o and hypopnea, oxygen
w 6
saturation and pulse. The
0.2 Time (min)
statistical index results are 0
0.1 0 5 10 15
shown in Fig. 11, which Fig. 8 Extracted
0 gives an obvious advantage respiratory parameters: 20 25 30
0 2 4 RR, MV, TV, and PIF.
8 10 in understanding symptom 35 40 45
12 of patient and thereby 50
R making a quick and accurate T
e diagnosis for OSA. i
s m
p e
i
r o
a
f
t
o
r d
y u
r
R a
a t
t i
e o
20
RR (bpm)

n
10
0 (
0 2 4 6 8 10 12
0.02 nute Ventilation s
)
0.04 Mi 96
(L)
MV(%)

465
SpO2
SpO2 00
Distribution of
40 SpO2
2000

Quantity
94 0
0 2 4 95 95.5 96 96.5
8 10
12 97.5 98 98.5 99
SpO2 (%)
Pulse
1 Rate Distribution of
2000
Pulse (bpm)

0 pulse rate

Quantity
0
1000
9
0
8
0
70 0
0 2 4 78 80 82 84
8 10 86 88 90
12 92 94 96
Ti
me
Pulse rate (bpm)
(mi
n)

Fig. 9 Extracted oxygen saturation of blood and pulse rate.

analyses of vital signs.

466
 IV. CONCLUSIONS
A BSN-based portable device to monitor respiratory
parameters using a micro thermal flow sensor to monitor
respiratory airflow, a tri-axis micro accelerometer to monitor
body posture, and a micro photoelectric sensor to monitor
blood oxygen saturation is developed. Homemade micro hot-
film flow sensor, extremely sensitive to weak airflow
(threshold of 0.01L/min), be capable of detecting infants’
respiration, is used to detect respiratory flow. The monitor
system utilizes Bluetooth communication, uses a cellphone or
PDA as the near terminal for tentative diagnosis, and uses
mobile cellular networks and Internet to transmit data to
remote medical center and physicians for professionally
analyzing, assessing, and diagnosing. The algorithms for data
processing and extracting vital signs of OSA patients are
designed and experimentally tested.
ACKNOWLEDGMENT
This work was supported by the National High-tech Program
‘863’ of China under the Grant 2012AA02A604.
REFERENCES
[1] A. P. Abidoye, N. A. Azeez, A. O. Adesina, K. K. Agbele1, and H. O.
Nyongesa, “Using wearable sensors for remote healthcare monitoring
system”, Journal of Sensor Technology, vol. 1, pp. 22-28, 2011.
[2] M. Rofouei, M. Sinclair, R. Bittner, T. Blank, and J. Heffron, “A non-
invasive wearable neck-cuff system for real-time sleep monitoring”,
2011 International Conference on Body Sensor Networks, pp.156-161,
2011.
[3] Z. Cao, R. Zhu, and R. Y. Que, “A Wireless portable system with micro
sensors for monitoring respiratory diseases”, IEEE Transactions on
Biomedical Engineering, vol. 59, issue 11, pp 3110-3116, 2012.
[4] R.Y. Que, and R. Zhu, “Aircraft aerodynamic parameter detection using
micro hot-film flow”, Sensors, vol. 12, pp.10920-10929, 2012.
[5] P. Liu, R. Zhu, “A flexible flow sensor system and its characteristics for
fluid mechanics measurements”, Sensors, vol. 9, pp. 9533-9543, 2009.
[6] R. Y. Que, R. Zhu, Q. Z. Wei, and Z. Cao, “Temperature compensation
for thermal anemometers using temperature sensors independent of flow
sensors”, Measurement Science and Technology, vol. 22, pp. 085404,
2011.
[7] "Sleep-related breathing disorders in adults: recommendations for
syndrome definition and measurement techniques in clinical research”.
The Report of an American Academy of Sleep Medicine Task Force.
Sleep, vol. 22, issue 5, pp. 667–89. August 1999.
[8] S. Oh, and H. Kwon, “Ubiquitous health monitoring system for diagnosis
of sleep apnea with Zigbee network and wireless LAN”, Journal of
Nanotechnology in Engineering and Medicine, vol. 2, pp. 021008, 2011.