2238

IEEE SENSORS JOURNAL, VOL. 15, NO. 4, APRIL 2015

Personal Lung Function Monitoring Devices

for Asthma Patients
Alice M. Kwan , Alexander G. Fung, Peter A. Jansen, Michael Schivo, Nicholas J. Kenyon,
Jean-Pierre Delplanque, and Cristina E. Davis

Abstract Asthma affects over 300 million people worldwide.

Asthmatics experience difficulty in breathing and airflow
obstruction caused by inflammation and constriction of the
airways. Home monitoring of lung function is the preferred
course of action to give physicians and asthma patients a chance
to control the disease jointly. Thus, it is important to develop
accurate and efficient asthma monitoring devices that are easy
for patients to use. While classic spirometry is currently the
best way to capture a complete picture of airflow obstruction
and lung function, the machines are bulky and generally
require supervision. Portable peak flow meters are available
but are inconvenient to use. There also exist no portable
inexpensive exhaled breath biomarker devices commercially
available to simultaneously measure concentrations of multiple
chemical biomarkers. We have created a user-friendly, accurate,
and portable external mobile device accessory that collects
spirometry, peak expiratory flow, exhaled nitric oxide, carbon
monoxide, and oxygen concentration information from patients
after two breath maneuvers. We have also developed a software
application that records and stores the gathered test information
and e-mails the results to a physician. Telemetric capabilities
help physicians to track asthma symptoms and lung function
over time, which allow physicians the opportunity to make
appropriate changes in a patients medication regimen more
quickly.
Index Terms Spirometry, peak expiratory flow (PEF),
chemical biomarker, personal mobile devices, smart devices,
breath analysis.

Manuscript received December 18, 2013; revised May 30, 2013 and October 10, 2014; accepted November 7, 2014. Date of publication November 24,
2014; date of current version February 5, 2015. This work was generously
and partially supported by several funding agencies. The content of this
work is solely the responsibility of the authors and does not necessarily
represent the official view of these agencies. Partial support is acknowledged
from: UL1 RR024146 #TR00002 from the National Center for Research
Resources (NCRR), a component of the National Institutes of Health (NIH),
and NIH Roadmap for Medical Research [JPD, CED, NJK]; NIH #HL
105573 [NJK]; The Hartwell Foundation [CED, NJK, JPD]; NIH #T32HL007013 and #T32-ES007059 [MS]; UC Davis School of Medicine and
NIH #8KL2TR000134-07 K12 mentored training award [MS]; American
Association for University Women (AAUW) Selected Professions Fellowship
support [AMK]. The associate editor coordinating the review of this paper
and approving it for publication was Dr. Patrick Ruther. ( Alice M. Kwan
and Alexander G. Fung contributed equally to this work.) (Corresponding
author: Cristina E. Davis.)
A. M. Kwan, A. G. Fung, J.-P. Delplanque, and C. E. Davis are with
the Department of Mechanical and Aerospace Engineering, University of
California at Davis, Davis, CA 95616 USA (e-mail: cedavis@ucdavis.edu).
P. A. Jansen is with the Product Development, Scanadu, Inc., Moffett Field,
CA 94035 USA.
M. Schivo and N. J. Kenyon are with the Division of Pulmonary and Critical
Care Medicine, University of California at Davis, Davis, CA 95616 USA.
Digital Object Identifier 10.1109/JSEN.2014.2373134

I. I NTRODUCTION
STHMA is a chronic pulmonary inflammatory disease
that affects the airways, and is characterized by
an increased sensitivity to various stimuli. Subsequent
stimulation may prompt the airways to narrow and induce
production of mucus causing less air to flow into the lungs.
Common symptoms of asthma include wheezing, shortness of
breath, and chest tightness. The intensity of an acute asthma
exacerbation, also known as an asthma attack, is unpredictable
and has the potential to be life threatening. While there are
medical treatments available to alleviate asthma symptoms,
there is no cure [1].
As of 2004, approximately 300 million people worldwide
were afflicted with asthma [2]. In 2010, 25.7 million
individuals were estimated to have asthma in the United
States [3]. Complications due to asthma accounted for
1.7 million emergency room visits in the United States
in 2006 [4], about 14.2 million lost work days in adults
in 2008, and annual total cost to society of nearly $56 billion
dollars [5]. More than 5 million children have asthma and the
prevalence of asthma is greater than 15% for children living
in low-income families in the United States [6].
The severity of symptoms, triggers, and responsiveness to
treatment medication are often unique to each individual.
Thus, a comprehensive guideline for an asthma action plan
recommends focusing on monitoring asthma symptoms as
a goal for asthma therapy [7]. Spirometry, peak expiratory
flow measurement, and a non-invasive marker of airway
inflammation known as fractional exhaled nitric oxide (FeNO)
are now used by health care professionals for diagnosis and
monitoring [8].
A spirometry test is a physiological test normally performed
under the supervision of trained professionals. It measures the
volume and flow rate of air that can be inhaled and exhaled,
and is useful in describing the disease state in the lungs, assessing therapeutic intervention, and/or monitoring for adverse
reactions to medication. Two of the most important parameters
obtained in a spirometry test are the forced vital capacity (FVC), described as the volume delivered during expiration
when made as forcefully and completely as possible starting
from full inspiration, and the forced expiratory volume in one
second (FEV1 ), which is the volume delivered in the first second of the FVC maneuver [9]. Prior published work has shown
that the forced expiratory volume in six seconds (FEV6 ) taken
during a spirometry maneuver is an acceptable substitute for

1530-437X 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

KWAN et al.: PERSONAL LUNG FUNCTION MONITORING DEVICES FOR ASTHMA PATIENTS

FVC [10]. The results of a spirometry test are presented using

a spirometry graph which maps flow rate (L/s) over exhaled
volume (L).
Peak expiratory flow (PEF) has been established as an accurate, repeatable, and non-invasive test for monitoring airflow
at home [11,12]. PEF is the maximum flow rate of expiration,
which correlates to the degree of obstruction in the airways.
Prescribed asthma monitoring plans with PEF monitoring
have been shown to decrease the number of severe asthma
episodes [13], but are ineffective when not adhered to [14,15].
This noncompliance may be due to the time and discipline
required to manually assess asthma symptoms over a long
time frame [16]. In addition, biomarkers like nitric oxide (NO)
and carbon monoxide (CO) can give health care professionals
another potential tool to help evaluate and determine asthma
treatment [17].
Portable technology focusing on full-body physiological
monitoring using sensors and mobile devices is becoming
increasingly prevalent. Recent research in this area has focused
on monitoring various physiological functions such as sweat
rates [18], cardiac function [19], sleep [20], and biomarkers
produced during exercise [21]. These devices provide patients
with an inexpensive, portable, and convenient method to noninvasively measure biomarkers of body function to improve
healthcare.
The aim of this paper is to report a novel portable approach
to asthma monitoring. An asthma monitoring device was
developed that combines spirometry (FEV1 , FEV6 , and
spirometry graph), PEF, and chemical breath biomarker
measurements of nitric oxide (NO), carbon monoxide (CO),
and oxygen (O2 ) into two breath maneuvers. A software
application for Android mobile technology was developed
to interpret and display relevant data, record the data to
a file on the Android device, and e-mail the data file
to a health care professional for personalized care. This
paper outlines the development of the instrumentation and
software application that enables portable and inexpensive
real-time collection of lung function parameters. Future
versions of these platforms may be particularly appropriate
for pediatric patients who may have greater difficulty
documenting their asthma symptoms during the course of
a day.
II. M ATERIALS AND M ETHODS
A. Design and System Construction of the Portable
Asthma Monitoring Device
1) Overview of the Device Layout: A novel portable asthma
monitoring device was created to provide the capability to
detect several critical lung function parameters and record
the data to a mobile device (Fig. 1). Patients exhale into
a flow chamber with embedded sensors that are tethered
to a smart device for data capture. A microcontroller and
USB host shield are used to digitize and send the sensor
signals to a mobile device through a standard USB connection.
A custom software application on the mobile device processes
the signal to communicate relevant physiological information
back to the patient and allow the processed data to be easily
shared via telemetry.

2239

Fig. 1. (a) Schematic overview of the asthma monitoring device and its components. (b) The portable asthma monitoring device with an Android Motorola
Xoom tablet. (c) The hardware components of the asthma monitoring device.

2) Flow Chamber Specifications: To measure the flow rate

of exhaled breath, a thin plate orifice volume flow meter [22]
was tailored for the expected pulmonary flow rates (Fig. 2(a)).
This type of volume flow meter allows an instantaneous
flow rate measurement in a pipe and follows the Bernoulli
obstruction theory, which describes the properties of a flow
that is forced by an obstruction from a duct of diameter D
into a smaller flow passage of diameter d. A relationship
between the pressure drop and volume flow rate is obtained
from Bernoullis equation:

2( p1 p2 )
(1)
Q = Co A o
(1 4 )
d
(2)
=
D

2240

Fig. 2. (a) Diagram of the flow chamber, (b) Computer-aided design drawing
of the flow chamber and its dimensions.

where Ao is the area of the hole in the orifice plate, p1 is

the pressure upstream of the orifice plate, p2 is the pressure
downstream of the orifice plate, is the density of exhaled
breath, is the diameter ratio as shown in Equation 2, and
Co is the orifice discharge coefficient which is typically on the
order of 0.6. The orifice discharge coefficient is a function of
the Reynolds number and [23]. The variable h is the length
of the orifice plate between diameters D and d. The chemical
sensors were placed at least 8 times the value of h away
from the orifice plate to ensure that exhaled breath would flow
over the chemical sensors [24]. In the device, the parameter
values were: D = 28 mm, d = 14 mm, h = 7 mm, and
P1, P2, A, B, and C indicate the location of pressure sensors,
NO sensor, CO sensor, and O2 sensor, respectively (Fig. 2(a)).
The flow chamber was constructed out of acrylonitrile butadiene styrene (ABS) using a Stratasys FDM rapid prototyping
machine (GoEngineer; Santa Clara, CA) (Fig. 2(b)) and was
designed to meet the requirements set by the American Society
of Mechanical Engineers (ASME) [25].
With these requirements, the flow meter theoretically
operates within a Reynolds number range of 104 107
(turbulent regime). For the physical size dimension in our flow
chamber, we find that flow rates from 300 L/min to 1000 L/min
have Reynolds numbers on the order of 104 .
3) Selection of Sensors: PEF values vary depending on
gender, age, and height of an individual [26]. In healthy
adult women, the maximum flow rate is approximately
450500 L/min while in healthy adult men, the maximum flow

IEEE SENSORS JOURNAL, VOL. 15, NO. 4, APRIL 2015

rate observed is approximately 600650 L/min [26]. Low flow

rate in this study was considered to be exhaled flow rate that
is at or below tidal breathing, which has been reported to be
around 41 11 L/min in healthy adults [27].
Therefore, to fully capture the dynamic range of exhaled
breath, two piezoresistive pressure sensors were selected to
monitor high flows of 50900 L/min (pressure sensor A;
model #MPX5010; Freescale Semiconductor; San Jose, CA),
and low flows of 15100 L/min (pressure sensor B; model
#SSCSNBN002NDAA5; Honeywell; Morristown, NJ).
The dynamic ranges for three of the chemical
biomarkers found in exhaled breath in asthma patients are
0.020.13 parts per million (ppm) for NO [28-31], 27 ppm
for CO [32, 33], and 1420 parts per hundred (pph) for
O2 [34] whereas in healthy individuals the dynamic ranges are
0.0050.02 ppm for NO, 12.3 ppm for CO, and 1420 pph
for O2 [31, 34]. The chemical sensors were selected to
detect the lower end of the biomarker concentration range
found in exhaled breath in asthma patients (model numbers
NO-D4, CO-D4, and O2-G2; AlphaSense Ltd.; Essex,
United Kingdom).
These three sensors are electrochemical sensors. The
oxygen sensor has a slight humidity dependence while the
NO and CO sensors do not have a humidity dependence
but have signal spikes from rapid transient changes in
humidity [35, 36]. The NO and CO sensors are rated for 80%
of the original signal after 2 years while the oxygen sensor
is rated for 85% of the original signal after 2 years [37-39].
For the NO and CO sensors, a potentiostatic circuit was built
to control the chemical sensor and a transimpedence amplifier
was used to convert the current generated from sensors to
a measureable voltage. The O2 sensor does not require a
potentiostatic circuit and the signal was obtained by using a
transimpedence amplifier to convert the current generated by
the sensor into a measureable voltage.
Quantification of chemical biomarkers in exhaled breath
must also occur before spirometry maneuvers because
spirometry often causes exhaled NO concentrations to
artificially decrease [40].
4) Microcontroller and USB Host Shield: A microcontroller
(Arduino UNO, R2; Strambino, Italy) was paired with a
USB host shield (Sparkfun, DEV-09947; Boulder, CO) to
control the external sensors and transmit data back to the
smart device. The Arduino UNO and USB host shield system
sends the digital signal in a three byte message from the
microcontroller to the Android mobile device using a USB
connection. During the first breath maneuver, the microcontroller collects data from the three chemical sensors first at
a rate of 100 samples/s for a total of 15s (data from each
chemical sensor is recorded for five seconds in a sequential
manner). In a second breath maneuver for spirometry testing,
data points were collected from pressure sensors A and B at a
rate of 50 samples/s. The microcontroller alternates sampling
between each pressure sensor which occurs for a total of 18 s,
allowing ample time for the patient to perform the spirometry
maneuver.
5) Software Application Development: An Android
software application for Android devices with the Gingerbread

KWAN et al.: PERSONAL LUNG FUNCTION MONITORING DEVICES FOR ASTHMA PATIENTS

2241

Each data point received from the microcontroller is

interpreted using linear regression equations determined from
the subsequent calibration experiments. All sensor data, time,
date, and global positioning system (GPS) location are written
in a comma separated value (.csv) file. Only the average
NO, CO, and O2 concentrations, PEF, instantaneous flow
rate, and the spirometry graph are shown to the patient on the
Android tablet screen (Fig. 3). These data can be e-mailed
using the native operating system e-mail application on the
tablet.
The software application prompts the patient to perform
70 seconds of tidal breathing to collect exhaled biomarker data.
Patients have five seconds of rest before the device signals
the patient to perform a full spirometry breath maneuver
(exhalation for at least 6 seconds to obtain a suitable FVC
alternate and an acceptable spirometry maneuver). PEF values
were identified from the spirometry maneuver.
To create a spirometry graph, flow rate measurements
from each pressure sensor are recorded in separate arrays.
After the maneuver has been completed, data from both
pressure sensors are merged. Data from pressure sensor B
are recorded if the flow rates from pressure sensor A are
less than 100 L/min. Otherwise, data from pressure sensor A
are recorded. The start of the test is determined using the
back extrapolation method recommended by Miller, et al. [9]
and the new time-zero denotes the start time for all timed
measurements. The volume flow rate time-series is integrated
numerically using a second-order method to produce a
corresponding exhaled-volume time series needed to plot the
spirometry loop and evaluate FEV1 and FEV6 .
Data obtained by each chemical sensor received from the
microcontroller is converted to a concentration value when
received by the software application. A numerical array was
created for each chemical biomarker and all data received after
the fifth time constant found for each sensor was averaged to
obtain an estimate of the concentration of the three chemical
biomarkers in exhaled breath.

Fig. 3.

Flowchart of the Android software application program.

platform (version 2.3.3) was created to extract important data

values from the microcontroller and to provide an interface
for the user (Fig. 3). Development was completed using
a Motorola Xoom tablet. The application was developed
in the Eclipse Indigo Integrated Development Environment
(version 3.7.1) and written in Java programming language
(version 1.6).

B. Calibration and Validation of Hardware

System Performance
1) Pressure Sensor Calibration: Compressed air and
a digital mass flow controller (Omega FMA 5545-ST;
Stamford, CT), were used for calibration. The voltage
output signal response from pressure sensor A was recorded
for flow rates ranging from 0500 L/min in 50 L/min
increments (n = 5). For pressure sensor B, the same setup
was used but was calibrated from 0100 L/min in 10 L/min
increments (n = 5).
2) Chemical Sensors Calibration: Three tanks of
compressed gas (1 ppm NO, 10 ppm CO, and 15 pph O2 )
were purchased from Airgas, Inc. (Sacramento, CA) and
diluted to concentrations found in asthma patients using gas
proportioners. The calibration was performed five times for
each chemical sensor at each concentration. Specifically, the
concentrations of NO gas used to calibrate the NO sensor
were: 0, 0.033 ppm, 0.05 ppm, 0.067 ppm, 0.1 ppm,
0.125 ppm, 0.2 ppm, 0.33 ppm, 0.5 ppm, and 1 ppm. The
concentrations of CO gas used to calibrate the CO sensor

2242

Fig. 4.

IEEE SENSORS JOURNAL, VOL. 15, NO. 4, APRIL 2015

(a) Pressure sensor A calibration. (b) Pressure sensor B calibration.

were: 0 ppm, 1 ppm, 1.67 ppm, 2 ppm, 2.5 ppm, 3.33 ppm,
5 ppm, 7 ppm, and 10 ppm. The concentrations of O2 gas used
to calibrate the O2 sensor were: 3 pph, 3.3 pph, 3.7 pph, 5 pph,
7.5 pph, 8.8 pph, 10 pph, 12.5 pph, 15 pph, and 20.9 pph.
3) Estimation of the Chemical Sensor Noise Floor: The
noise floor of each sensor was calculated according to
Equation 3,
NoiseFloor = x + 3

(3)

where x represents the mean of the signals collected in ambient

air and is the standard deviation. The intersection between
the linear regression line and the noise floor is marked as the
limit of detection (LOD). The LOD indicated for each sensor
signifies the minimum concentration of analyte distinguishable
from the background noise of the sensor.
4) Time Constant Determination: Each chemical sensor was
specified by the vendor to have a response time to 90% of
the final concentration of less than 25 seconds. Determination
of the time constant identified the requisite time for tidal
breathing. At five times the time constant, the sensors will
reach 99.3% of the final concentration. The change in chemical
concentration from ambient conditions to the breath sample
produces a step response which can be characterized for a
first-order system by equation 4.
1
y = 1 exp( t)

(4)

where y is the percentage of the final value, t is the time in

seconds since the step, and is the time constant. Given that
the time to 90% of the final value is known, equation 4 can
be rearranged to solve for the time constant.

Fig. 5. (a) Nitric oxide sensor calibration. (b) Carbon monoxide sensor
calibration. (c) Oxygen sensor calibration.

III. R ESULTS
A. Calibration of Pressure Sensors
Both pressure sensors were calibrated to correlate sensor
output with flow rate through the device. The flow equation
(Eq. 1) shows that the pressure drop across the obstruction
flow meter is proportional to the square of the flow rate. The
correlation equations needed to convert the voltage output, V,
of each pressure sensor into a flow rate value Q (L/min) were
constructed accordingly: V = k Q 2 + k (with k = A, B).
Linear regression yielded the values of the parameters for each
sensor: A = 7 106 V/(L/min)2 , A = 1.83 102 V
(R 2 = 0.996) and B = 1.11 104 V/(L/min)2 ,
B = 2.567 V (R 2 = 0.989) as shown in Fig. 4(a) and (b).
These correlation equations were coded into the software
application. With calibration, pressure sensor A is able
to accurately measure flow rates as low as 50 L/min
(14.4% error). Pressure sensor B enables the measurement of
flow rates as low as 15 L/min with a 5.517% error.
B. Chemical Sensors and Time Constants
All three chemical sensors exhibited a linear relationship
(V = k C + k with k = NO, CO, O2 ) between analyte
concentration, C, and sensor output voltage, V (Fig. 5(a)-(c)).

KWAN et al.: PERSONAL LUNG FUNCTION MONITORING DEVICES FOR ASTHMA PATIENTS

They were calibrated to establish the values of the

regression parameters for each sensor: NO = 0.514V/ppm,
NO = 0.176V (R 2 = 0.999); CO = 0.325V/ppm,
CO = 0.732V (R 2 = 0.999); O2 = 0.141V/pph,
O2 = 0.660V (R 2 = 0.995) for NO, CO, and O2 respectively.
The limits of detection are (Fig. 5(a)-(c)): LODNO =
0.029 ppm (0.190 V noise floor), LODCO = 0.945 ppm
(0.1038 V noise floor), and LODO2 = 4.722 pph (1.308 V
noise floor). The NO LOD is very close to the low-end values
of NO concentrations expected for asthma patients (0.03 ppm).
This indicates that while the NO sensor is theoretically suitable
to measure NO in exhaled breath at such low concentrations,
the output values may not be reliable for concentrations at
or lower than 0.03 ppm. The CO LOD and O2 LOD are
respectively lower than the lowest CO concentrations expected
in the breath of asthma patients (2 ppm) and well below the
lowest O2 concentration in breath (14 pph).
The chemical sensor response times are provided as t90 in
their data sheets which are <15, <20, and <25 seconds for
NO, O2 , and CO respectively. Using equation 4, this results in
time constants of 6.51, 8.69, and 10.86 seconds for NO, O2 ,
and CO respectively. Thus, the patient needs to breath for at
least 55 seconds in order for the sensors to reach their final
value.
C. Chemical Sensor Performance Using Gas Mixtures
The performance of each chemical sensor was verified to
confirm their ability to measure the concentration of a target
gas in mixture with other two gases of interest (e.g. NO along
with CO, and O2 ) with acceptable precision and accuracy.
Specific concentrations were achieved by controlling the flow
rate ratios of each gas. The three gases were mixed using
two gas proportioners to produce three distinct mixtures that
contained concentrations of NO, CO, O2 within the range
found in the exhaled breath of asthma patients. A standard
procedure was used to prepare each mixture. Three tanks of
specialty gas were used for this experiment (1 ppm NO tank,
1000 ppm CO tank, and 15 pph O2 tank). NO and O2 were
first mixed in a gas proportioner with a flow rate ratio of
1 NO:X O2 . This NO/O2 mixture was then sent to another
gas proportioner where it was mixed with CO. The flow rate
ratio for the second gas proportioner was 1 CO:Y NO/O2 .
Because the concentration of exhaled O2 should be similar
for all individuals, all gas mixtures were made with an O2 concentration of 1314%. This oxygen concentration was chosen
to reflect the low end of the sensor range in order to confirm
that the sensors could operate the lowest O2 concentration in
human breath.
For NO and CO, the concentrations were varied between
the upper and lower limits of the concentrations in the
exhaled breath of asthma patients. The first mixture contained
NO and CO concentrations at the upper limit of the concentration range of those with asthma and O2 at the low end
of the concentration range in human breath (0.099 ppm NO,
6.944 ppm CO, and 13.406 pph O2 ; X = 9 and Y = 143).
The second mixture contained NO and CO concentrations
near the middle of the concentration range of those with

2243

TABLE I
C HEMICAL S ENSOR R ESPONSE TO G AS M IXTURES OF
N ITRIC O XIDE , C ARBON M ONOXIDE , O XYGEN

asthma and O2 at the low end of the concentration range

in human breath (0.066 ppm NO, 4.975 ppm CO, and
13.930 pph O2 ; X = 14 and Y = 200). The third mixture
contained NO and CO concentrations at the low end of
the concentration range of those with asthma and O2 at
the low end of the concentration range in human breath
(0.033 ppm NO, 1.996 ppm CO, and 14.471 pph O2 ; X = 29
and Y = 500).
Each gas sensor was exposed to the three gas mixtures
five times and the resulting concentration readings were
recorded (Table I). A two-tailed t-test was used to compare the
measured and theoretical gas concentrations. For the mixtures,
only oxygen was significantly different, which indicates that
the oxygen sensor likely needs to be recalibrated.
D. Spirometry Accuracy and Repeatability
Prior to performing a prospective validation study with
asthmatic and non-asthmatic subjects, we believed it was
critical to test the performance and accuracy of this asthma
monitoring device in the laboratory setting. Performance
was assessed and compared against a spirometer currently
approved for use in a clinical setting (SDI Diagnostics,
SPIROLABII).
A group of 10 subjects was asked to perform three
acceptable spirometry maneuvers using the clinical
spirometer under the supervision of a health care professional.
An acceptable maneuver is described as a full, uninterrupted
exhalation of breath that was not interrupted (coughing,
hesitation, etc.), had satisfactory exhalation duration (6 s),
and resulted from maximum exhalation effort by the patient.
The same subjects were then asked to perform three more
acceptable spirometry maneuvers using the asthma monitoring
device. Only forced exhalation is assessed with the asthma
monitoring device, thus only PEF, FEV1 , and FEV6 .

2244

IEEE SENSORS JOURNAL, VOL. 15, NO. 4, APRIL 2015

TABLE II
C OMPARISON OF THE A STHMA M ONITORING D EVICE
W ITH A C LINICAL S PIROMETER

Fig. 6. (a) Ideal inhalation and exhalation flow-volume spirometry graph from
a theoretical normal healthy adult. (b) Exhalation flow-volume spirometry
graph generated by a subject using the asthma monitoring device.

Repeatability of the asthma monitoring device was

evaluated in accordance with the requirements established by
the American Thoracic Society and the European Respiratory
Society [9], where the two largest values of FEV1 and FEV6
must be within 0.150 L. The 10 test subjects were able to
perform three acceptable spirometry maneuvers on the clinical
spirometer before performing three acceptable spirometry
maneuvers on the asthma monitoring device without difficulty.
During the spirometry maneuver, the flow rate increases at
a steep positive slope until the peak flow is reached,
approximately one second from the start of exhalation. The
flow rate then gradually decreases until the total lung volume
has been exhaled, approximately 6 seconds from the start of
exhalation. The volume of exhaled breath is plotted on the
x-axis of the spirometry graph. At the end of the 6 seconds of
exhalation, the flow rate reaches zero and the corresponding
x-axis value would be considered the total lung capacity for
that subject (Fig. 6(a) and 6(b)). For each subject, the averages

of the three values of PEF, FEV1 , and FEV6 recorded by

the asthma monitoring device were compared against the
averages of the same parameters taken from the clinical
spirometer using 2-block analysis of variance (ANOVA)
with one block covering the variation amongst people and
the other covering the variation between the devices. The
collected data is summarized in Table II and Fig. 7(a)-(c).
While there was no significant difference between the
devices for PEF and FEV1 measurements (F(1,9) = 2.381,

KWAN et al.: PERSONAL LUNG FUNCTION MONITORING DEVICES FOR ASTHMA PATIENTS

Fig. 7. (a) Average PEF values from all subjects using the asthma monitoring
device and a clinical spirometer. (b) Average FEV1 values from all subjects
using the asthma monitoring device and a clinical spirometer. (c) Average
FEV6 values from all subjects using the asthma monitoring device and a
clinical spirometer.

p = 0.157 and F(1,9) = 0.653, p = 0.440 respectively),

there was a significant difference in the FEV6 measurements
(F(1,9) = 38.541, p = 0.00016).
IV. D ISCUSSION
This study presents a new approach to monitoring lung
function in asthma patients with a novel portable device
that operates using a smart phone or tablet. This device
allows the acquisition of the expiration branch of a spirometry
test and the associated parameters (i.e. PEF, FEV1 , FEV6 ),
as well as the quantification of relevant exhaled biomarkers
(NO, CO). Initial testing of this asthma monitoring device
in the laboratory setting has demonstrated its capability to
measuring major lung function parameters with reasonable
accuracy and precision.
The flow metering function of the device is performed
using an obstruction flow meter equipped with two differential

2245

pressure sensors, each focusing on half of the expected volume

flow rate range. High flow rates are generated at the beginning
of the spirometry maneuver and pressure sensor A was needed
to ensure that the device captured PEF values accurately. There
was no significant difference between the PEF value measured
by a clinical spirometer and that obtained with the asthma
monitoring device. Hence, pressure sensor A effectively measures PEF. Pressure sensor B was selected to capture the low
flow rates (<50 L/min) that are characteristic at the end of a
spirometry maneuver. While pressure sensor B does measure
flow rates lower than tidal breathing effectively, it cannot
accurately measure flow rates that are below 15 L/min. This
limitation may result in an underestimation of FEV6 as the
device does not quite capture the full spirometry maneuver.
Accuracy in the low flow rate range could be improved without
changing the overall design concept by implementing a more
sensitive and higher cost, pressure sensor.
A ten-sample test was conducted in the research setting
to determine if the asthma monitoring device can perform
a spirometry test comparable to the expiration branch of a
clinical spirometer. As noted above, the PEF values acquired
using the asthma monitoring device are not significantly
different from the clinical spirometer. After several forced
expiratory maneuvers, patients may experience fatigue and
be inclined to stop exhalation before they have completed
the maneuver. Spirometry measurements are still considered
acceptable if the drop in FEV1 or FEV6 does not exceed
20% if more than three maneuvers are required [9, 41] which
is the case for all measurements of the 10 subjects. Further
testing is needed to determine if the asthma monitoring device
can accurately detect lung function parameters in those who
regularly smoke tobacco. The significant differences in FEV6
values were expected because of the limitations in pressure
sensor B, which does not allow the asthma monitoring
device to accurately detect flow rates lower than 15 L/min.
Furthermore FEV6 values measured are often dependent on
the effort the patient puts into fully completing the spirometry
maneuver, which can be inconsistent.
The variability in measured spirometry lung function
parameters from the asthma monitoring device and the clinical
spirometer were observed to be very similar. For PEF values,
the standard deviation of measurements taken on the asthma
monitoring device ranged between 5.11956.102 L/min,
while the clinical spirometer had a standard deviation
range of 5.57483.302 L/min. The standard deviation of
FEV1 measurements taken on the asthma monitoring device
was 0.0230.333 L, and the clinical spirometer had a standard
deviation of FEV1 measurements of 0.0150.405 L. The standard deviation of FEV6 measurements from the asthma monitoring device was between 0.0710.373 L, and the clinical
spirometer had standard deviations of FEV6 values between
0.0100.260 L. Aside from illustrating the inherent variability
of spirometry measurements, this data also shows that 9 out of
the 10 subjects had their PEF and FEV1 measurements from
both devices overlap with one another. This along with the lack
of a significant difference indicates that the asthma monitoring
device can measure these parameters with an accuracy and
precision comparable to that of the clinical spirometer.

2246

The chemical sensors used in this work (AlphaSense, Ltd.)

were selected for their combination of high sensitivity, high
selectivity, miniature size, and short response time compared
to other available chemical sensors. They were shown to have
a linear relationship between voltage and concentration of
analyte and the CO and NO chemical sensors were able to
detect various concentrations of their target analytes accurately
within a mixture (no significant difference). The difference
in the oxygen sensor is likely due to a miscalibration which
can easily be corrected. The time constants determined for the
chemical sensors ranged from 6.51 seconds for the NO sensor
to 10.86 seconds for the CO sensor. Patients therefore need to
perform tidal breathing, exhaling through their mouth, for at
least 70 seconds (55 for the sensors and 15 to read the values)
to ensure that the chemical biomarkers in their breath can be
read. This duration might be difficult for severely asthmatic
patients and future design improvements should incorporate
chemical sensors with a shorter response time and better sensitivity to the target analyte. From experimentation it was seen
that the NO LOD (0.029 ppm NO) was very close to the lower
end of the dynamic range found in asthma patients (0.03 ppm).
This would normally indicate that at concentrations as low as
0.03 ppm NO, the sensor output signal could be confused with
the background noise of the sensor. However, as seen in the
experiment using three gas mixtures, the NO sensor was able
to detect a 0.03 ppm NO concentration with 3.727% error.
Although the lower limit of 0.03 ppm is above the lower
bound of the American Thoracic Society (ATS) guidelines
(0.025 ppm for adults and 0.02 ppm for children), it is below
the lower bound of the group that is likely to benefit from
medication (>0.05 ppm for adults and >0.35 ppb for children)
[31]. This range would indicate that this device would be of
particular benefit to those beginning medication.
A key feature of the asthma monitoring device described
here is its ability to measure multiple lung function parameters
through two breath maneuvers. Though there is a number of
personal PEF, spirometers, and NO sensing devices currently
on the market, a device that combines PEF measurements,
spirometry, and NO, CO, and O2 biomarker detection is
new. One of the advantages of being able to measure PEF,
spirometry, and exhaled chemical biomarkers at once on a
smart phone or tablet is the opportunity to utilize the built-in
connectivity features such as GPS and e-mail. The ability
to gather the necessary data quickly and efficiently and then
instantly communicate that data with a health care professional
means that such devices have the potential to significantly
improve the speed of respiratory health care and asthma
management in the future.
Given the ubiquity of advanced mobile devices today,
designing personal medical monitors to be compatible with
smart phones and tablets is an effective way to help lower
development costs and increase user compliance. An integral
part of ensuring that patients have the tools to control and
manage their own asthma symptoms is to provide patients
with an accurate, reliable, and portable device that can be
accessed at any time. A major problem associated with current
handheld PEF devices available to patients is that values
must be written and recorded in a daily symptoms journal.

IEEE SENSORS JOURNAL, VOL. 15, NO. 4, APRIL 2015

The goal of the portable nature of our asthma monitoring

device and its ability to connect with Android mobile devices
is to encourage asthma patients to play a more active role
in controlling their disease. Another important advantage of
our device is its capability of saving each test onto the
Android tablet and e-mailing the recorded data, which gives
health care professionals a chance to study the symptoms
and lung function patterns of a patient over time. These
patterns may shed light on trends regarding asthma triggers
and the effectiveness of new medications and treatments. The
asthma monitoring device detailed in this paper represents
a simple solution to improve asthma monitoring compliance
by providing a portable, inexpensive, and user-friendly option
that gives both patients and doctors a complete snapshot of
lung health.
V. C ONCLUSION
We have designed, built, and validated a personalized lung
function monitoring device that utilizes smart phone and tablet
technology to create a convenient, reliable, and user-friendly
system. Initial validation testing has proved that measurements
taken with this device are comparable to that of a clinical
spirometer and satisfy the minimum requirements for spirometry as outlined by Miller and colleagues. Advancements
toward personalized medicine provide more opportunities to
perform longitudinal studies with asthma patients remotely
and enable patients to become more aware of their lung health.
ACKNOWLEDGMENTS
The authors would also like to thank Dr. M. Delwiche
(Biological and Agricultural Engineering; University of
California, Davis) for instrumentation advice, and Scanadu
Inc. for interest in the project.
R EFERENCES
[1] (2012). What is Asthma? [Online]. Available: http://www.nhlbi.nih.
gov/health/health-topics/topics/asthma/
[2] M. Masoli, D. Fabian, S. Holt, R. Beasley, and Global Initiative for
Asthma (GINA) Program, The global burden of asthma: Executive
summary of the GINA dissemination committee report, Allergy, vol. 59,
no. 5, pp. 469478, May 2004.
[3] L. J. Akinbami et al., Trends in asthma prevalence, health care use, and
mortality in the United States, 2001-2010, National Center for Health
Statistics, Hyattsville, MD, USA, NCHS data brief, no. 94, 2012.
[4] S. R. Pitts, R. W. Niska, J. Xu, and C. W. Burt, National hospital
ambulatory medical care survey: 2006 emergency department summary, National Health Statistics, Hyattsville, MD, USA, Tech. Rep. 7,
Aug. 2008, pp. 138.
[5] S. B. L. Barnett and T. A. Nurmagambetov, Costs of asthma in
the United States: 20022007, J. Allergy Clin. Immunol., vol. 127,
pp. 145152, Jan. 2011.
[6] B. Bloom, R. A. Cohen, and G. Freeman, Summary health statistics
for U.S. children: National health interview survey, 2010, Nat. Center
Health Statist. Vital Health Statist., vol 10, no. 250, pp. 180, Dec. 2011.
[7] National Asthma Education and Prevention Program, Expert panel
report 3 (EPR-3): Guidelines for the diagnosis and management of
asthma-summary report 2007, J. Allergy Clin. Immunol., vol. 120,
pp. S94S138, Nov. 2007.
[8] E. D. Bateman et al., Global strategy for asthma management and
prevention: GINA executive summary, Eur. Respirat. J., vol. 31, no. 21,
pp. 143178, 2008.
[9] M. R. Miller et al., Standardisation of spirometry, Eur. Respirat. J.,
vol. 26, no. 2, pp. 319338, Aug. 2005.

KWAN et al.: PERSONAL LUNG FUNCTION MONITORING DEVICES FOR ASTHMA PATIENTS

[10] M. P. Swanney, R. L. Jensen, D. A. Crichton, L. E. Beckert,

L. A. Cardno, and R. O. Crapo, FEV6 is an acceptable surrogate for
FVC in the spirometric diagnosis of airway obstruction and restriction,
Amer. J. Respirat. Critical Care Med., vol. 162, no. 3, pp. 917919,
2000.
[11] J. M. Ignacio-Garcia and P. Gonzalez-Santos, Asthma self-management
education program by home monitoring of peak expiratory flow, Amer.
J. Respirat. Critical Care Med., vol. 151, no. 2, pp. 353359, Feb. 1995.
[12] M. R. Miller, O. F. Pedersen, and P. H. Quanjer, The rise and dwell time
for peak expiratory flow in patients with and without airflow limitation,
Amer. J. Respirat. Critical Care Med., vol. 158, no. 1, pp. 2327, 1998.
[13] P. G. Gibson et al., Self-management education and regular practitioner
review for adults with asthma, Cochrane Database Syst. Rev., vol. 3,
no. 3, p. CD001117, 2003.
[14] P. V. Burkhart, M. K. Rayen, W. R. Revelette, and A. Ohlmann,
Improved health outcomes with peak flow monitoring for children with
asthma, J. Asthma, vol. 44, no. 2, pp. 137142, 2007.
[15] A. W. A. Kamps, R. J. Roorda, and P. L. P. Brand, Peak flow diaries
in childhood asthma are unreliable, Thorax, vol. 56, pp. 180182,
Mar. 2001.
[16] J. Ct, A. Cartier, J.-L. Malo, M. Rouleau, and L.-P. Boulet, Compliance with peak expiratory flow monitoring in home management of
asthma, Chest, vol. 113, no. 4, pp. 968972, 1998.
[17] M. Maniscalco and J. O. Lundberg, Hand-held nitric oxide sensor
NIOX MINO for the monitoring of respiratory disorders, Expert Rev.
Respirat. Med., vol. 4, no. 6, pp. 715721, 2010.
[18] P. Salvo, F. Di Francesco, D. Constano, C. Ferrari, M. G. Trivella,
and D. De Rossi, A wearable sensor for measuring sweat rate, IEEE
Sensors J., vol. 10, no. 10, pp. 15571558, Oct. 2010.
[19] L. Hadievski et al., A novel mobile transtelephonic system with
synthesized 12-lead ECG, IEEE Trans. Inf. Technol. Biomed., vol. 8,
no. 4, pp. 428438, Dec. 2004.
[20] K. Watanabe, Y. Kurihara, T. Nakamura, and H. Tanaka, Design
of a low-frequency microphone for mobile phones and its application to
ubiquitous medical and healthcare monitoring, IEEE Sensors J., vol. 10,
no. 5, pp. 934941, May 2010.
[21] G. W. Hunter et al., Smart sensor systems for human health breath
monitoring applications, J. Breath Res., vol. 5, no. 3, p. 037111,
Sep. 2011.
[22] B. R. Munson, D. F. Young, and T. H. Okiishi, Fundamentals of Fluid
Mechanics, 5th ed. New York, NY, USA: Wiley, 2006.
[23] F. White, Fluid Mechanics, 6th ed. New York, NY, USA: McGraw-Hill,
2008.
[24] A. P. Singh, A. R. Paul, and P. Ranjan, Investigation of reattachment
length for a turbulent flow over a backward facing step for different step
angle, Int. J. Eng., Sci., Technol., vol. 3, no. 2, pp. 8488, 2011.
[25] J. L. Hankinson, J. R. Odencrantz, and K. B. Fedan, Spirometric
reference values from a sample of the general U.S. population, Amer.
J. Respirat. Critical Care Med., vol. 159, no. 1, pp. 179187, Jan. 1999.
[26] I. Gregg and A. J. Nunn, Peak expiratory flow in normal subjects,
Brit. Med. J., vol. 3, no. 5874, pp. 282284, 1973.
[27] M. J. Morris and D. J. Lane, Tidal expiratory flow patterns in airflow
obstruction, Thorax, vol. 36, pp. 135142, Feb. 1981.
[28] E. Baraldi, N. M. Azzolin, S. Zanconato, C. Dario, and F. Zacchello,
Corticosteroids decrease exhaled nitric oxide in children with acute
asthma, J. Pediatrics, vol. 131, no. 3, pp. 381385, Sep. 1997.
[29] C. A. Byrnes, S. Dinarevic, E. A. Shinebourne, P. J. Barnes, and A. Bush,
Exhaled nitric oxide measurements in normal and asthmatic children,
Pediatric Pulmonol., vol. 24, no. 5, pp. 312318, 1997.
[30] M. Fleischer et al., Detection of volatile compounds correlated
to human diseases through breath analysis with chemical sensors,
Sens. Actuators B, Chem., vol. 83, nos. 13, pp. 245249, 2002.
[31] R. A. Dweik et al., An official ATS clinical practice guideline: Interpretation of exhaled nitric oxide levels (FeNO) for clinical applications,
Amer. J. Respirat. Critical Care Med., vol. 184, no. 5, pp. 602615,
2011.
[32] C. G. Uasuf, A. Jatakanon, A. James, S. A. Kharitonov, N. M. Wilson,
and P. J. Barnes, Exhaled carbon monoxide in childhood asthma,
J. Pediatrics, vol. 135, no. 5, pp. 569574, 1999.
[33] K. Zayasu, K. Sekizawa, S. Okinaga, M. Yamaya, T. Ohrui, and
H. Sasaki, Increased carbon monoxide in exhaled air of asthmatic
patients, Amer. J. Respirat. Critical Care Med., vol. 156, no. 4,
pp. 11401143, 1997.
[34] V. Wenzel, A. H. Idris, M. J. Banner, R. S. Fuerst, and K. J. Tucker,
The composition of gas given by mouth-to-mouth ventilation during
CPR, Chest, vol. 106, no. 6, pp. 18061810, Dec. 1994.

2247

[35] How Humidity Affects Oxygen Sensor Output, Alphasense Ltd.,

Great Notley, U.K., pp. 12.
[36] Design Considerations in Gas Sampling, Alphasense Ltd., Great Notley,
U.K., pp. 16.
[37] O2-G2 Oxygen Sensor, Alphasense Ltd., Great Notley, U.K., May 2013.
[38] CO-D4 Carbon Monoxide Sensor, Alphasense Ltd., Great Notley, U.K.,
Jan. 2012.
[39] NO-D4 Nitric Oxide Sensor, Alphasense Ltd., Great Notley, U.K.,
Oct. 2013.
[40] P. E. Silkoff et al., Exhaled nitric oxide after 2 -agonist inhalation and
spirometry in asthma, Amer. J. Respirat. Critical Care Med., vol. 159,
no. 3, pp. 940944, 1999.
[41] S. Stanojevic et al., Reference ranges for spirometry across all ages:
A new approach, Amer. J. Respirat. Critical Care Med., vol. 177, no. 3,
pp. 253260, Feb. 2008.

Alice M. Kwan received the B.S. degree in biomedical engineering from the
University of California at San Diego, La Jolla, CA, in 2008 and the M.S.
degree in mechanical engineering from the University of California at Davis,
Davis, CA, USA, in 2012. She is currently working in the medical device
industry.

Alexander G. Fung is currently pursuing the Ph.D. degree in mechanical

engineering at the University of California at Davis, Davis, CA, USA, where
he received the B.S. degree in mechanical engineering, in 2011.

Peter A. Jansen is currently a Post-Doctoral Research Fellow with the

University of Arizona, Tucson, AZ, USA. He was with Scanadu, Inc., as
a Senior AI Engineer and received the Ph.D. degree in neural computation
and cognitive language from McMaster University, Hamilton, ON, Canada,
in 2010.

Michael Schivo is currently with the Reversible Obstructive Airway Diseases

Center, the UC Asthma Network Clinics, and the Center for Comparative
Respiratory Biology and Medicine, University of California at Davis, Davis,
CA, USA. He also has a general pulmonology clinic once a week in Rocklin,
CA, USA. He collaborates highly with biologists, chemists, chemometricians,
engineers, and pediatric/adult clinicians to coordinate clinical trials utilizing
VOC analysis in disease states. His research spans from in vitro epithelial
models to human subject studies.

Nicholas J. Kenyon is currently a Professor with the University of California

at Davis (UC Davis), Davis, CA, USA. His translational research program
focuses on severe asthma and COPD, and the role of nitric oxide in airway
inflammation. He joined the faculty of UC Davis in 2001.

Jean-Pierre Delplanque is currently a Professor with the University of

California at Davis (UC Davis), Davis, CA, USA, where his research program
focuses on theoretical and computational fluid dynamics and transport phenomena. His specific applications include inert and reactive multiphase flow
systems, and emerging materials processing methodologies and processes.
He has been with UC Davis since 2004.

Cristina E. Davis is currently a Professor with the University of California

at Davis (UC Davis), Davis, CA, USA, where her research program focuses
on design and implementation of sensors and instrumentation for biological
applications. She has been with UC Davis since 2005.