[ieee 2013 6th biomedical engineering international conference (bmeicon) - amphur muang, krabi,...

5
*corresponding author Automatic Fall Detection Sensor for Treadmill Rehabilitation Wongwit Senavongse*, Noramon Dron, Pornsuang Prakopkaew, Wanidar Tammawong Department of Biomedical Engineering, Faculty of Engineering Srinakharinwirot University, Ongkharak, Nakhonnayok, Thailand [email protected]* Abstract—Stroke patient rehabilitation by treadmill tends to increase in recent years. The rehabilitation on treadmill needs more than two staffs per patient to control the process and protect them from falling. As a result, it takes too much time and resource to operate. To resolve this problem, this paper proposes to develop an automatic sensor that is simple and easy to use by using ultrasonic sensor and the Arduino board. The system senses patient’s body on treadmill to calculate the suitable distance for setting an automatic switch. When patient’ body triggers the first position, the treadmill will start moving then patient’s body triggers the second distance or fall position, the treadmill will stop immediately. Although it is a simple concept, the commercial treadmill does not have this function which is necessary for looking after patients. The control system uses Arduino and C# languages. The findings suggest that the performance of the system is very satisfactory having accuracy which are measured in 2 states, on and off, at 94.32% and 94.83% respectively. Keywords—treadmill; rehabilitation; stroke; automatic sensor; fall detection; ultrasonic; walking I. INTRODUCTION World Stroke Organization: WSO reports stroke is the second leading cause of death of people aged more than 60 years old in the World population. More than 6,000,000 people died each year from stroke and following stroke attack, 50% of the survivors will have some significant disabilities [1] especially gait impairments. Some patients never able to walk and others have a permanent residual loss of walking ability. Treadmill retraining has emerged as a potential treatment in this decade with many researches supported the treatment [2]. The treadmill is a device for walking while staying in the same place. The machine provides a moving platform with a wide conveyor belt driven by an electric motor, the belt moves to the rear with constant speed. In general hospital, the walking rehabilitation uses a low cost treadmill with no suitable safety system. It has only emergency stop that requires the user intervention. Thus, one patient may need 2-4 staff supports for controlling treadmill and looking after to protect any accident that may occur during the rehabilitative program. The specific treadmill with safety supported system has a very high cost. Most hospitals cannot or choose not to effort it. The objective of this study is to implement the rehabilitation treadmill using a minimal staff for patient, with low cost but having the same or much performance. Therefore, a low cost automatic sensor was developed to cooperate with an ordinary treadmill. Its function can automatically turn the treadmill on and off, and detect patient’s body. If the patient falls or cannot keep pace with, the treadmill will stop immediately. II. MATERIAL AND METHOD A. AUTOMATIC FALL DETECTION SENSOR ALGORITHM This sensor is based on ultrasonic sensor and Arduino board. Ultrasonic sensors measure the distance using the echo principle. The piezo element generates a sound wave and measures the time required for the piezo to transmit and then receive the sound wave. The travel time is the basis for measuring the distance to the target [3] using the following equation: Distance to Object = (T x Speed of sound)/2 (1) T = Time between an ultrasonic wave is emitted and it is received. Division by 2 is because the sound wave has to travel to the object and back [4]. The ultrasonic sensor will sense the patient’s body and send the signal through Arduino board to control the treadmill. Fig. 1. How sensor operates treadmill According to Figure 1, patient’s body is detected to calculate two suitable distances. When patient’s body triggers the first distance causing the treadmill to move him back slowly and patient’s body subsequently triggers the second distance causing the Treadmill to gradually stop. When the patient is ready, he will step towards the ultrasonic sensor and the cycle repeats. If he falls on the treadmill, a buzzer will 978-1-4799-1467-8/13/$31.00 ©2013 IEEE The 2013 Biomedical Engineering International Conference (BMEiCON-2013)

Upload: wanidar

Post on 07-Mar-2017

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: [IEEE 2013 6th Biomedical Engineering International Conference (BMEiCON) - Amphur Muang, Krabi, Thailand (2013.10.23-2013.10.25)] The 6th 2013 Biomedical Engineering International

*corresponding author

Automatic Fall Detection Sensor for Treadmill Rehabilitation

Wongwit Senavongse*, Noramon Dron, Pornsuang Prakopkaew, Wanidar Tammawong Department of Biomedical Engineering, Faculty of Engineering

Srinakharinwirot University, Ongkharak, Nakhonnayok, Thailand [email protected]*

Abstract—Stroke patient rehabilitation by treadmill tends to increase in recent years. The rehabilitation on treadmill needs more than two staffs per patient to control the process and protect them from falling. As a result, it takes too much time and resource to operate. To resolve this problem, this paper proposes to develop an automatic sensor that is simple and easy to use by using ultrasonic sensor and the Arduino board. The system senses patient’s body on treadmill to calculate the suitable distance for setting an automatic switch. When patient’ body triggers the first position, the treadmill will start moving then patient’s body triggers the second distance or fall position, the treadmill will stop immediately. Although it is a simple concept, the commercial treadmill does not have this function which is necessary for looking after patients. The control system uses Arduino and C# languages. The findings suggest that the performance of the system is very satisfactory having accuracy which are measured in 2 states, on and off, at 94.32% and 94.83% respectively.

Keywords—treadmill; rehabilitation; stroke; automatic sensor; fall detection; ultrasonic; walking

I. INTRODUCTION

World Stroke Organization: WSO reports stroke is the second leading cause of death of people aged more than 60 years old in the World population. More than 6,000,000 people died each year from stroke and following stroke attack, 50% of the survivors will have some significant disabilities [1] especially gait impairments. Some patients never able to walk and others have a permanent residual loss of walking ability.

Treadmill retraining has emerged as a potential treatment in this decade with many researches supported the treatment [2]. The treadmill is a device for walking while staying in the same place. The machine provides a moving platform with a wide conveyor belt driven by an electric motor, the belt moves to the rear with constant speed.

In general hospital, the walking rehabilitation uses a low cost treadmill with no suitable safety system. It has only emergency stop that requires the user intervention. Thus, one patient may need 2-4 staff supports for controlling treadmill and looking after to protect any accident that may occur during the rehabilitative program. The specific treadmill with safety supported system has a very high cost. Most hospitals cannot or choose not to effort it. The objective of this study is to implement the rehabilitation treadmill using a minimal staff for patient, with low cost but having the same or much

performance. Therefore, a low cost automatic sensor was developed to cooperate with an ordinary treadmill. Its function can automatically turn the treadmill on and off, and detect patient’s body. If the patient falls or cannot keep pace with, the treadmill will stop immediately.

II. MATERIAL AND METHOD

A. AUTOMATIC FALL DETECTION SENSOR ALGORITHM This sensor is based on ultrasonic sensor and Arduino

board. Ultrasonic sensors measure the distance using the echo principle. The piezo element generates a sound wave and measures the time required for the piezo to transmit and then receive the sound wave. The travel time is the basis for measuring the distance to the target [3] using the following equation:

Distance to Object = (T x Speed of sound)/2 (1)

T = Time between an ultrasonic wave is emitted and it is received. Division by 2 is because the sound wave has to travel to the object and back [4]. The ultrasonic sensor will sense the patient’s body and send the signal through Arduino board to control the treadmill.

Fig. 1. How sensor operates treadmill

According to Figure 1, patient’s body is detected to calculate two suitable distances. When patient’s body triggers the first distance causing the treadmill to move him back slowly and patient’s body subsequently triggers the second distance causing the Treadmill to gradually stop. When the patient is ready, he will step towards the ultrasonic sensor and the cycle repeats. If he falls on the treadmill, a buzzer will

978-1-4799-1467-8/13/$31.00 ©2013 IEEE

The 2013 Biomedical Engineering International Conference (BMEiCON-2013)

Page 2: [IEEE 2013 6th Biomedical Engineering International Conference (BMEiCON) - Amphur Muang, Krabi, Thailand (2013.10.23-2013.10.25)] The 6th 2013 Biomedical Engineering International

alarm and the treadmill will stop immediately. This process is shown in Figure 1.

Fig. 2. Block diagram

An ultrasonic sensor, max sonar EZ3, is used to detect the distance between the sensor and patient’s body. The distance data is sent to Arduino board which determines and processes the distance data received. If the patient is ready on the treadmill, the program in Arduino board will send the data signal to Computer through C# programming to control the treadmill to turn on and off automatically. If the patient’s body cannot be detected, it means the patient falls. This will cause the program on Arduino board to make the buzzer alarm and the specific code is sent to the PC through C# program to stop the treadmill immediately. The sensor used in this study is explained in the following sections.

B. Hardware Design Aduino board ET-BASE AVR EASY88 was used in the

part of buzzer alarm connecting buzzer at SPK(D3) port. The buzzer was connected through transister BC 337 as show in the Figure 3.

Figure. 3 Part of buzzer circuit

For the ultrasonic sensor, the sensor was connected through Arduino analog0 PC0 port and another wire for Vcc and ground. When everything is set for the distance value of ultrasonic sensor, the signal will be sent to ATmega88 in Arduino board as shown in Figure 3. The analog voltage from ultrasonic sensor has direct variation to the distance of the patient.Ultrasonic sensor output is scaled to the input power provided

to the sensor. The formula for the voltage scaling on the ultrasonic sensor is shown in equation (2):

[(Vcc/512) = Vi] (2)

Vcc = Supplied Voltage, and Vi = Volts per inch (Scaling)

The microprocessor is used to calculate the distance of the patient from ultrasonic sensor signal.

Figure 4. Part of ultrasonic sensor circuit

Having connected computer pc through RS232 as USB port, the data code can be sent C# programming to command the treadmill. The circuit of USB port to connect to computer is shown in Figure 5.

Figure 5. Part of USB port circuit

C. Software Design

Software design consists of two parts of programming, the first is Arduino programming for receiving data from ultrasonic sensor and the second is C# programming for controlling treadmill.

ARDUINO

Page 3: [IEEE 2013 6th Biomedical Engineering International Conference (BMEiCON) - Amphur Muang, Krabi, Thailand (2013.10.23-2013.10.25)] The 6th 2013 Biomedical Engineering International

The distance between sensor’s pole and the beginning of treadmill’s platform was measured to specify the first standard distance value that commands the treadmill to start moving. Another distance from sensor’s pole to the second distance was measured as the safety distance before the end of treadmill’s platform to specify as the standard value. All distance information is gathered and the flow chart of the program was written as shown in Figure 6.

Figure 6. Flow chart1 Arduino programming

The program begins to work as the sensor detects the distance of the patient on the treadmill. If the patient walks straight and triggers at the first distance that has a value equal or less than 75. If it is yes, Arduino board will send the start code A to computer C# to command the treadmill to start. When the patient moves back and triggers the second distance that has a value equal or more than 120, the Arduino board will send the stop code B to computer C# to command the treadmill to stop. However, if the sensor detects a value more than 200, it implies that the patient falls or has an accident. The signal will be sent to trigger buzzer alarm and also stop the treadmill. The first and second distance can be adjusted to be appropriate to any treadmill.

C# Program

The template C# program was written to receive the code from Arduino board in order to determine whether to turn the treadmill on or off.

Figure 7. Flow chart2 C# programming

The flow chart of C# program is shown in Figure 7 to demonstrate both conditions either sending start code to treadmill if receive ‘A’ from Arduino board or sending stop code if it receive ‘B’. The start and stop code sent to treadmill depend on the different brands and models of treadmill.

Once the hardware and software parts are complete, the experiments are carried out to measure the accuracy of the system. The experiments are performed to find an error occurring when treadmill is turned on and off to make sure that patient will be safe while operating the treadmill.

To test the performance of the sensor, the hp cosmos treadmill was used in the experiment. Ten subjects were used to test the sensor with different factors that may affect the accuracy of the sensor such as body shape, height, chest width and gender. The standard distance between sensor’s pole and the beginning of treadmill’s platform was set as 37.5 inch with the other standard distance from sensor’s pole to second distance setting at 60 inch. The data were collected 5 times per person and the average of first distance (distance1) and second distance (distance2) were calculated by using the following equation:

Mean = ( X)/N (3)

The protocol of the test is to first let the subject walk on the treadmill when the treadmill is turned on, then the first distance is measured. Next, the subject moves back slowly to trigger the second distance and treadmill stops. The second distance is then measured.

Page 4: [IEEE 2013 6th Biomedical Engineering International Conference (BMEiCON) - Amphur Muang, Krabi, Thailand (2013.10.23-2013.10.25)] The 6th 2013 Biomedical Engineering International

III. RESULTS AND DISCUSSIONA low cost automatic sensor was designed so that it can

cooperate with an ordinary treadmill as shown in Figure 8.

(A) (B)

Figure 8. Automatic Fall Detection Sensor (A) and sensor on treadmill (B)

The findings from the test of performance of the sensor are shown in the Table I.

TABLE I. ACCURACY FOR FALL DETECTION SENSOR

TABLE I the results of the accuracy of the sensor.

The means of the accuracy are 93.26% and 94.83% respectively for accuracy1 and accuracy2. The accuracy1 is the accuracy of first distance whereas the accuracy2 is the accuracy of second distance. The accuracy was found using the equation(4),

Accuracy = (Average distance/Standard distance) x 100 (4)

Base on the experiment of accuracy measurement, the highest accuracy of first distance is 100% for female subject2 and the lowest accuracy is 89.33% for male subject9. This is because most subjects stepped across the standard distance, that makes the distance become shorter. However, the system still works well to prevent the fall of the patient so it is justified to say that this is not a problem as the safety is still properly maintained.

The highest accuracy of second distance is 98.33% for male subject8, and the lowest accuracy is 90.5% for female subject3. This is because the shape of body. Looking at the

subject data from table I, most males provide more accuracy than female. That may be explained by the anatomy and physiology of most males in this study who have flat and wide chest but females have more curved and contoured of their bodies. Therefore, male can reflect the ultrasonic wave better than female which cause the treadmill to stop with more delays in female because of the scattered reflection. Even though there is an error, the distance set for the treadmill to stop is still sufficient to stop before patient move to the rear part of the treadmill or to stop when the sensor cannot detect the patient. There is a limitation that this sensor has a little time delay in female as mentioned above but it always stops right before the subject moves out of treadmill. In this way, this sensor is sufficiently safe for the patient to operate the treadmill and the safety is ensured in the treadmill for any gender. For real application, this sensor system can be developed for personal rehabilitation equipment. Therefore, every criterion should be optimized for a particular user for the best performance of system.

IV. CONCLUSIONIn this paper, an automatic fall detection sensor was

developed for treadmill rehabilitation to be low cost and automatic. The sensor can also be incorporated with an ordinary treadmill using ultrasonic sensor to protect stroke and spinal cord injury patient while they do rehabilitation on the treadmill. This will be of a great benefit to health care as nursing staff to look after the patients can be minimized to save cost and time. If the patient falls during the rehabilitation the program will stop treadmill immediately so nurse staff can rush to help the patient without the concern over the running treadmill. The findings of this study show that the performance of the sensor is very satisfactory having the mean accuracy of 93.26% and 94.83%, respectively for accuracy1 and accuracy2. It can be concluded that the automatic fall detection sensor for treadmill has a useful function which can reduce nurse staffs, can be made at low cost, can offer good safety and can be used with any treadmill. This treadmill sensor can now be used for patient rehabilitation and can be developed further to add speed control function to improve exercise mode and improve the accuracy.

REFERENCES

[1] http://www.strokecenter.org/patients/about-stroke/stroke-statistics/ [2] Manning, C D and Pomeroy, V M (2003). ‘Effectiveness of treadmill

retraining on gait of hemiparetic stroke patients: Systematic review of current evidence’, Physiotherapy, 89, 6, 337-349.

[3] http://www.ece.uvic.ca/~elec499/2002b/group13/Theory.html[4] http://www.education.rec.ri.cmu.edu/products/nxt_video_trainer2/resour

ces/helpers/nxt_sensors/ultrasonic.html[5] Kendrick, C, Holt, R, McGlashan, K, Jenner, J R and Kirker, S

(2001).‘Exercising on a treadmill to improve functional mobility in chronic stroke: Case report’, Physiotherapy, 87, 5, 261-265.

[6] S. Crompton, M. Khemlani1, J. Batty, L. Ada, C. Dean and P. Katrak, “Practical issues in retraining walking in severely disabled patients using

Page 5: [IEEE 2013 6th Biomedical Engineering International Conference (BMEiCON) - Amphur Muang, Krabi, Thailand (2013.10.23-2013.10.25)] The 6th 2013 Biomedical Engineering International

treadmill and harness support systems”, Australian Journal of Physiotherapy, Vol. 47, 2001.

[7] A.L. Hicks, PhD, K.A. Martin Ginis, PhD, “Treadmill training after spinal cord injury: It’s not just about the walking”, JRRD, Volume 45, Number 2, Pages 241–248, 2008

[8] Harkema SJ, Hillyer J,Schmidt-read M, Ardolino D, Sisto SA, Behrman AL, “Locomotor training: as a treadment of spinal cord injury and in the progession of neurologic rehabilitation”, Arch Phys Med Rehabil, 2012; 93: 1588-97.

[9] Wass E, Taylor NF, Matsas A, “Familiarisation to treadmill walking in unimpaired older people”, E. Wass et al. / Gait and Posture 21 p.72–79, 2005.

[10] E. B. Macbight, M. R. Popovic, nd T. Adam Thrasher, “Functional Electrical Therapy for Assisted Treadmill Training: Use of Passive

Dynamics”, Proceedings of the 25' Annual Intematianal Conference of the IEEE EMBS Cansun, Mexico - September 17-21,2003.

[11] Aaslund MK, Helbostad JL, Moe-Nilssen R, “Familiarisation to body weight supported treadmill training for patients post-stroke”, M.K. Aaslund et al. / Gait & Posture 34, p.467–472, 2011.

[12] ZHU Su-li1,2 LONG Li-rong1, “The Treadmill Effect on the Utility of Quality of Working Life”, International Conference on Management Science & Engineering, 2008.

[13] David Juang, M.D., Frankie B. Fike, M.D., Carrie A. Laituri, M.D., Vincent E. Mortellaro, M.D., and Shawn D. St. Peter, M.D., “ASSOCIATION FOR ACADEMIC SURGERY Treadmill Injuries in the Pediatric Population”, Journal of Surgical Research 170, p.139–142, 2011.