A Smartphone-based Posture Measurement System for Physical

Therapy Applications

Synchronization of Multiple Devices via Bluetooth Network

Noriaki Ikeda

, Kai Ishida

, Yoshitaka Shiba

, Kousuke Mizuno

Noritaka Mamorita

and Akihiro Takeuchi

Medical Informatics, School of AHS, Kitasato University, Sagamihara, Kanagawa, Japan

Graduate School of Medical Sciences, Kitasato University, Sagamihara, Kanagawa, Japan

Physical Therapy, School of AHS, Kitasato University, Sagamihara, Kanagawa, Japan

Division of Rehabilitation, Kitasato University East Hospital, Sagamihara, Kanagawa, Japan

Department of Clinical and Rehabilitation Engineering, Hokkaido Institute of Technology, Sapporo, Japan

Keywords: Bluetooth, iphone/ipod, Posture Measurement, Physical Therapy.

Abstract: A smartphone-based measurement system was developed for the purpose of measuring time series data for

posture in the physical therapy field. We used the iPod touch (Apple Inc.) as hardware and iOS SDK as a

software development tool. Posture data (pitch, roll and yaw) were taken directly from Euler angles or by

transformation from quaternion data (qw, qx, qy, qz) to the Euler angles, depending on the orientation of the

device. This approach allows continuity of data values. Data were stored in the Documents directory of the

iOS Appli as a file in CSV format, which can be transferred to a PC via iTunes or sent by email as an

attached file if a WiFi environment is available. In order to synchronize two devices, communication via

Bluetooth was implemented. The accuracy of the data was checked by comparing with the OPTOTRAK

data. Variation of posture while standing still and walking was recorded using this system for 50 elderly

subjects.

1 INTRODUCTION

1.1 Background

The rapid aging of society in Japan has increased the

importance of approaches that allow elderly people

to live a healthy and independent life, in order to

reduce medical costs and to lighten the burden of

care.

Elderly people suffer various reductions in

physical and vital functions, including decreased

posture stability that reduces balance and the quality

of walking. In turn, these changes reduce ADL

(Activities of Daily Living) performance and QOL

(Quality of Life). Displacement of the pelvis has

been related to falling in elderly people (Ishigaki, et

al., 2011), which indicates the importance of posture

during walking, as well as while standing still.

Measurement of posture currently requires a

laboratory environment or a stationary position

sensor device (Ishigaki, et al., 2011). An easy-to-use

mobile device for this purpose is required for wider

application in community-dwelling elderly people.

1.2 Purpose of the Study

We have previously developed a system for

evaluating tremor symptoms in Parkinson’s disease

using the accelerometer installed in a game

controller (Mamorita, et. al., 2009). The purpose of

the present study is to develop a measurement

system for posture during walking that is compact,

easy to use, and inexpensive for physical therapy

applications.

2 METHODS

2.1 Posture Measurement System

As hardware, we chose the iPod touch (Apple Inc.,

Ikeda N., Ishida K., Shiba Y., Mizuno K., Mamorita N. and Takeuchi A..

A Smartphone-based Posture Measurement System for Physical Therapy Applications - Synchronization of Multiple Devices via Bluetooth Network.

DOI: 10.5220/0004203600800083

In Proceedings of the 2nd International Conference on Sensor Networks (SENSORNETS-2013), pages 80-83

ISBN: 978-989-8565-45-7

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

101 g, 58.9 × 111 × 7.2 mm) and the built-in sensors

of acceleration, rotation rate (Gyro sensor) and

attitude. Software was developed as an “iAppli”

using iOS SDK v.4.2 on an Apple Macintosh with

Mac OS X (ver 10.6.5).

The system measures 3-axis acceleration,

rotation rate and attitude. The coordinates of the

system (when the iPhone/iPod touch is laid flat on

the desk) are +Z, upward perpendicular; +X,

rightward (in portrait view); and +Y, upward (in

portrait view). We define three different modes of

the device as follows, because the data from the

attitude sensor is output to a different variable

depending on the placement of the device.

F (Flat): The device is placed flat with the +Z

axis pointing upward perpendicular.

H (Horizontal): The device is placed flat with the

+X axis pointing upward perpendicular.

V (Vertical): The device is placed flat with the -

Y axis pointing upward perpendicular.

The CMMotionManager class and the

CMAttitude class of CoreMotion Framework were

used to obtain the sensor data. Attitude data are

given as the Euler angles (pitch, roll and yaw),

quaternion data (qw, qx, qy, qz), and the rotation

matrix (m

, i,j=1,2,3) (Apple Inc., 2010). The

method of obtaining data from the sensor was

changed as follows, depending on the positioning

mode of the device.

F: Px=roll, Py= pitch, Pz= yaw, from the Euler

angles.

H: Same as F, but adding 90° to Pz.

V: Euler angles were calculated from qw, qx, qy,

and qz using the equations below, and 90° was

added to Py to make the reference value equal to

zero.

Px = atan2(2(qw*qx+qy*qz), 1-2*(qx*qx+qy*qy));

Py = asin(2*(qw*qy-qz*qx);

Pz = atan2(2*(qw*qz+qx*qy), 1-2*(qy*qy+qz*qz));

Data are stored in the “Documents” directory

as a CSV file with a file name constructed from the

date and time when the data were collected. These

files can be transferred to a PC via iTunes or sent by

email as an attached file if a WiFi environment is

available.

The display during measurement is shown in

Figure 1. Using pop-up menus, various functions can

be accessed, including setting the measuring time (5,

10, 20(default), 30, 40, 60 sec), reviewing recorded

data (up to 20 sec) with pinching in and out and

scrolling , display of statistics , display of file names

saved in the Documents directory, sending mail, and

display of the instruction manual. The functions of

connection by Bluetooth, selection of placement of

the device (F/H/V), and on/off for saving a file were

assigned to the buttons (see Figure 1).

If we want to measure the data from multiple

points at a time, inter-device synchronization is

required. This function was realized using Bluetooth

wireless communication with GameKit framework

(Apple Inc, 2010). Two devices are connected as

peer-to-peer mode by Bluetooth pairing procedure

(Figure 2).

2.2 Subjects and Measurements

The subjects in the study were 50 community-

dwelling elderly females from 6 districts of

Sagamihara city. The subjects had a mean age of

71.0 ± 4.9 (range: 61-82) years old, height of 152.3

± 5.3 cm, and body weight of 52.4 ± 7.0 kg. Each

subject answered questions on age, sex, height, body

weight, history of falling, and amount of daily

exercise. Measurements were made for physical

functions and activity, including muscle strength for

knee extension, bending angle of the upper body,

stimulus response time of the body, 10 m walking

time at a comfortable speed, and 10 m walking time

at maximum speed.

To evaluate the posture of the trunk, the device

was attached to the sternum with plastic tape so that

the +Y axis of the device was parallel to the

perpendicular direction (V position; Figure 3A). To

measure rotation of the pelvis, another device was

attached with a belt on the sacrum so that the +X

axis pointed in the upward perpendicular direction

(H position; Figure 3B).

The study was approved by the ethical

committee of Kitasato University School of

Medicine.

3 RESULTS

3.1 Evaluation of Posture Data

To evaluate the posture measurement, the results

were compared with the integral of gyro data. An

example is shown in Figure 4, where the blue, red

and green lines show the gyro (Gx), posture (Px),

and integral (Ix), respectively:

Ix(t) = ∫Gx(t) dt.

The absolute difference | Px(t) - Ix(t) | increased with

time (about 15.3° at t=20 s). However, if the device

A Smartphone-based Posture Measurement System for Physical Therapy Applications - Synchronization of Multiple

Devices via Bluetooth Network

was returned to the initial position, Px(t) showed

only a small error (about 0.5°) compared to the

initial value Px(0).

3.2 Comparison with the OPTOTRAK

To evaluate the accuracy of measured data we used

OPTOTRAK (Northern Digital Inc., Waterloo,

Canada) and compared the two data. One of the best

results is shown in Figure 5, in which the data from

OPTOTRAK (blue line) and those from iPod (green

one) were plotted. Time axes of two data were

adjusted (27 sample points were shifted) so that the

correlation between them was maximum (R=0.8886).

The RMS (root mean square of differences) value

was 2.15 (deg).

3.3 Example Data

In some cases we had observations that the yaw

component was subject to a linear drift (Figure 6)

especially in rapid movements. This phenomenon

did not appear in the other components (pitch and

roll) and the cause has not been identified yet.

4 DISCUSSION

A three-axis accelerometer is often used in walking

analyses in physical therapy (Hemmi et al., 2009);

(Kojima, et al., 2008), but a posture sensor has not

been available. Therefore, the device described in

this report is likely to be of value for many

applications, particularly since it can be used

outdoors, rather than being limited to the laboratory.

We used two devices to measure the movement

of the subject at two positions of the body. In order

to synchronize two devices (i.e. to adjust the time

axis of the two data files), onset of the data

acquisition was synchronized using Bluetooth

wireless communication. This connection, however,

was not so accurate nor stable; time to communicate

between two devices varied from 1 msec to several

10 msec. Considering that the sampling frequency

was 50 Hz, this is not satisfactory. We are trying to

do several methods to avoid this problem.

Three or more devices can be synchronized, but

the variance of delay will be larger. One can use

WiFi to synchronize remote multiple devices if the

wireless LAN environment is available, by which a

global sensor network could be constructed.

5 CONCLUSIONS

A smartphone-based measurement system was

developed for measuring time series data for posture

in physical therapy. Data for variation of posture on

standing still and during walking were recorded

using this system in 50 elderly subjects. The results

suggest that the system can serve as a new tool for

walking analysis in physical therapy.

ACKNOWLEDGEMENTS

This study was funded in part by a grant from

Kitasato University School of Allied Health

Sciences (No.2012-6604). The posture measurement

system can be downloaded from the App Store

(Ikeda, 2012).

REFERENCES

Apple Inc., 2010. CMAttitude Class Reference. iOS SDK

Reference Manuals.

Ishigaki, N., Kimura, T., Usui, Y. 2011. Analysis of pelvic

movement in the elderly during walking using a

posture monitoring system equipped with a triaxial

accelerometer and a gyroscope. Journal of

Biomechanics 44:9, 1788-1792.

Mamorita N., Iizuka T., Takeuchi A., Shirataka M., Ikeda

N., 2009. Development of a system for measurement

and analysis of tremor using a three-axis

accelerometer. Methods Inf Med 48: 589-594.

Hemmi, O., Shiba, Y., Saito, T., Tsuruta, H., Takeuchi, A.,

Shirataka, M., Obuchi, S., Kojima, M., Ikeda, N., 2009.

Spectral Analysis of Gait Variability of Stride Interval

Time Series: Comparison of young, elderly and

Parkinson’s disease patients. J. Phys. Ther. Sci.

21:105-111.

Kojima, M., Obuchi, S., Henmi, O., Ikeda, N., 2008.

Comparison of smoothness during gait between

community dwelling elderly fallers and non-fallers

using power spectrum entropy of acceleration time-

series J. Phys. Ther. Sci. 20:243-248.

Ikeda, N., 2012. GYRO-kun v3.0 (GR3e). AppStore.

URL: http://soul.ahs.kitasato-u.ac.jp/~ikeda/.

SENSORNETS 2013 - 2nd International Conference on Sensor Networks

APENDIX

Figure 1: Screen shot of the system. Buttons at the bottom

are, from left to right, Bluetooth connection, start/stop of

measurement, recording to file ON/OFF and the placement

of device (flat, horizontal or vertical).

Figure 2: Pairing of two iPods via Bluetooth.

Figure 3: Mounting of the device. A: Vertical setting to

measure the posture of the trunk. B: Horizontal setting to

measure the rotation of the pelvis.

Figure 4: Comparison of posture data (red) and integral

values (green) of rotation rate (blue).

Figure 5: Comparison of iPod data (green) and

OPTOTRAK data (blue).

Figure 6: An example of drift. The last half of yaw

component of posture data (red) measured by iPod was

subject to a drift.

A Smartphone-based Posture Measurement System for Physical Therapy Applications - Synchronization of Multiple

Devices via Bluetooth Network