Towards an Autonomous Fall Detection and Alerting System on a Mobile and Pervasive Environment Ivo C. Lopes2, Binod Vaidya1, and Joel J. P. C. Rodrigues1,2 1 2
Instituto de Telecomunicações, Portugal Department of Informatics, University of Beira Interior, Covilhã, Portugal
[email protected];
[email protected];
[email protected]
Abstract — In recent years, the use of sensors on mobile devices is highly desirable. In particular, an accelerometer can be used for numerous applications such as tracking object or monitoring of the elderly. This paper presents an application tool based on an accelerometer, call SensorFall to detect and report the acceleration caused by a fall, which allows sending alerts in the form of SMS, phone call, or by location using the GPS. We have implemented and verified the SensorFall in various environments, such as a hospital or a normal daily life for the elderly, also implemented the system calibration in order to adapt better the living conditions of each person. The results show that it performs well.
Keywords — mobile devices, sensors, accelerometer, fall detection
1
1. Introduction Every day new technologies appear and can be used in several applications. In recent days, incorporating sensors in the mobile devices such as mobile phones and portable digital displays (PDAs) is considered necessary. The sensor used is an accelerometer that can be applied for different uses and purposes, such as games, screen rotation, changing the music track, monitoring movements of the human body [1, 2], among others. Accelerometers are more and more being incorporated into personal electronic devices, like Smartphone, Digital Audio Players and PDAs have accelerometers for user interface control. There are several problems, which come with the falls, such as in hospitals and with elderly people. Falls and fall-induced injuries among older people are major public health concerns worldwide, accounting for over 80% of all injury-related admissions to hospital of people over 65 years [3, 4]. Falls are also the leading causes of unintentional injury death in these individuals and responsible for appreciable morbidity, including bone fracture, head injury, joint disruption, and soft tissue contusion and laceration resulting in pain, functional impairment, disability, fear of falling, depression, loss of independence and confidence, and admission to residential care [3, 5, 6]. This paper proposes a system that uses a regular mobile phone or PDA with an accelerometer to detect the acceleration caused by a fall and corresponding alert and report. The system provides a personal ubiquitous accessible tool on a PDA without the use of other extra devices, combining a regular PDA with an accelerometer, global system for mobile communications (GSM), and global positioning system (GPS) location for easy alerting and monitoring.
2
This system, named SensorFall, is an innovative detection system and assistance of falls on a mobile and pervasive environment, using an accelerometer to measure the accelerations caused by movement of the human body in the tasks of day-to-day, the system incorporates an array of features, such as sending alerts, or shortest message service (SMS), or phone calls or the GPS positions, another feature relates to the fact that it can be apply to anyone and is not only restricted to the elderly people, and can be calibrated and used by an athlete, for example. The aim of this project is to construct an application for mobile devices that allows the detection of falls and corresponding notification. It pretends to give fast assistance and quality of response, in assisting the patient or user. The system application monitors the acceleration in different axes X, Y and Z, in real time [7]. It can detect the change of the sensor and can decide if it is acceleration that is worthy of a fall or just a natural movement of labor of a dayto-day human being. When a fall happens, it will send a message notifying the user. After a few seconds, if a reply is not given, it will be issued a warning via SMS to a predefined mobile phone number. Its aim is to provide higher security to the user and to be an efficient mean of warning, such that hospital professionals can offer a prompt and efficient response and quick assistance. GPS is a positioning system that gives the geographical coordinates of any place on Earth, this position is referenced to the equator and the meridian of Greenwich and is reflected by three coordinates: latitude, longitude and altitude. It is now possible to have a global positioning system due to the use of satellites for a total of 24 satellites that orbit the Earth every 12 hours and frequently send radio signals. At each point of the Earth is always visible 4 satellites and the various signals of four satellites the GPS receiver to obtain latitude, longitude and altitude. Latitude is the distance from the equator measured along the Greenwich 3
meridian and the longitude is the distance from the meridian of Greenwich, both distances are measured in degrees. The system contains all the mechanisms to work with the GPS hardware of the PDA, and thus, whenever we're outdoors and an external connection using the wireless network or an mobile operator network, receive the values of latitude, longitude, number of satellites, ground level from the sea and hours, but in this paper only is necessary the data on latitude, longitude and number of satellites, which was good enough for the application. Another part important of our system is the calibration, where the user uses the device during a certain time, and carries out normal day-to-day for an elderly person to get the value have to be obviously different from an athlete who to run makes more acceleration, with this system is intended to extend a greater coverage, and above all, be more efficient and accurate. We propose the SensorFall because we believe that there is enormous potential in new technologies such as PDAs, which together with health is of tremendous help. There are several approaches and several implementations for the detection of falls, but none is used in one device (PDA), all features that our system has combined in one device that can be easily found on the market and reduced cost, all the necessary tools for the detection of falls. The rest of the paper is organized as follows. Section 2 reviews the related literature while Section 3 presents the system architecture, focusing on the application architecture, the system requirements, application development and technologies used to deploy the solution. The system is tested and its validation is performed in Section 4. Section 5 concludes the paper and point directions for further research works.
4
2. Related Work This section deals with some of the existing works related to the proposed mobile solution, mainly, using sensors for measuring and identifying different human behaviors. In [8], authors use acceleration signatures from everyday activities for on-body device location. They present a solution to favor the recognition of everyday life activities, through the changing positions that are detected using a set of sensors that are incorporated in phones, PDAs, and watches. These systems must identify the devices position in the body. Using a three-axis accelerometer, the authors describe two methods for measuring the time spent in vertical jumps [9]. These algorithms are based on the morphology of the signal. The first uses the maximum curve during the stage of landing, while the later uses the time interval between the minimum and maximum acceleration values during the flight (up) and landing (down), respectively. An approach based on wearable accelerometer sensors is presented in [10] for an ambulatory monitoring of human posture and walking speed. Authors propose a classification system for monitoring real-time physical activity, which will be able to detect body postures (lying, sitting and standing) and race. The system acquires data from a set of two-axis accelerometer implemented in a wireless body sensor network (BSN). A sensor is mounted on the waist, while other two are attached to their thighs. Uses only one sensor, does not incorporate the PDA, does not have alerts, compare with our proposal. An implementation of an accelerometer sensor module and fall detection monitoring system, based on wireless sensor networks, was proposed in [11]. The authors implement a wireless accelerometer to determine the posture of a person, 5
activity, and fall. The system uses a two-axis accelerometer (the ADXL202), and a wireless radio-frequency (RF) module. This module measures the signals from the accelerometer and shows them in a personal computer. Standing, sitting, lying down, walking and running are the activities detected by the accelerometer. Ours approach is better because it incorporates all the features of PDA, such as SMS, phone calls, accelerometers and GPS coordinates, in the detection of falls. Another approach for health monitoring using electrocardiogram (ECG) and three-axis accelerometers for elderly persons at home is presented in [12]. Authors present a prototype of "wellness" and surveillance system with capacity to record and analyze data from the ECG and the accelerometer. The resources of the electrocardiogram are used to detect risk of arrhythmias and so on. The system includes a base station, to collect the data from wireless sensors placed on the patient body. Walking and other body functions are activities detected and recorded by accelerometers. It uses the standard IEEE 802.15.4 protocol for communication between the sensors and the base station. If any abnormality occurs the system sends an alarm condition for a predefined cell phone. This system sends the alarm to a cell phone, but the system is not embedded in a PDA, such as our approach, isn’t as complete as the SensorFall. Authors of [13] propose a solution that tries to detect, in real time, the basic living activity at home using a wearable sensor and smart home sensors. This approach aims to locate people and detect their movements. Already in [14], a proposal based on a wearable accelerometer sensor was proposed for gait authentication and identification. In [15] and [16], different proposed methods but with a similar use of accelerometers, with different objectives, are presented. In [15], an accelerometer on the trunk of a person is used to recognized activities, such as lying, running, walking and standing. In [16], an accelerometer placed on the 6
trunk of a person tries to detect vibration noise caused by heartbeats and thus measure the heart. In [17], the authors introduce Fallarm, a pervasive fall prevention solution for hospitals and care facilities, as well as for home settings. They applied a versatile intervention strategy based on closed-loop information exchange between proactive and reactive methods, like, comprehensive assessment protocols determine the individuals' risk of falling, and an a device continuously monitors subjects' activities, and it provides patients with constant feedback about their actual risk. In none of these approaches the PDA is used, which differs from our system. In [18], the article is a survey of systems, algorithms and sensors, for the automatic early detection of the fall of elderly persons. It is only a theoretical system for the detection of falls. The authors of [19] propose a patient alert alarm for fall administration. It is ZigBee-Based location awareness fall detection system that provides instant position information to the caregivers as soon as the fall happen. At the [20] the authors suggest, a system to sense, send, display and store physiology activity. The system includes a wearable device to be worn by the individual to collect physical activity data, a wireless communication link between the user and the monitoring network, it is include fall detection and heart beat measurement. An approach using a three-axial accelerometer and barometric pressure measurement, presented in [21]. The authors present a falls detection system, employing a Bluetooth based wearable device, containing a triaxial accelerometer and a barometric pressure sensor, the plan of the paper is to evaluate the use of
7
barometric pressure measurement, as a surrogate measure of altitude, to augment previously reported accelerometry-based falls detection algorithms. In [22], the main results related to a fall detection system are shown by means of a personal server for the control and processing of the data acquired from multiple intelligent biomedical sensors. In [23], the authors present a novel fall detection system using both accelerometers and gyroscopes. They divide human activities into two categories: static postures and dynamic transitions. By using two tri-axial accelerometers at separate body locations, their system can recognize four kinds of static postures: standing, bending, sitting, and lying. Motions between these static postures are considered as dynamic transitions. Linear acceleration and angular velocity are measured to determine whether motion transitions are intentional. If the transition before a lying posture is not intentional, a fall event is detected. The authors of [24], present a system that detects human falls in the home environment, distinguishing them from competing noise, by using only the audio signal from a single far-field microphone. The proposed system models each fall or noise segment by means of a Gaussian mixture model (GMM) supervector, whose Euclidean distance measures the pairwise difference between audio segments. A support vector machine built on a kernel between GMM supervectors is employed to classify audio segments into falls and various types of noise. As in other approaches, the mentioned above, [19 - 24] do not use a PDA, such as our system uses, and allows a single unit has all the features. The approach [25], the authors embed a tri-axial accelerometer in a cell phone, connect to Internet via the wireless channel, and using SVM (Support Vector Machine) algorithm for the pre-processing. KFD (Kernel Fisher Discriminant), and k-NN (Nearest Neighbor) algorithm for the precise classification. This 8
approach uses a mobile phone, with an accelerometer applied, and it isn’t already incorporated in the PDA such as SensorFall, to raise the alert they uses wireless networks, but that implies that the user has Internet, in the case of older people do not happens. As it may be seen in the above-mentioned systems, most of them provide dedicate solutions using sensors to monitor different human behaviors. Our proposal is designed for a regular mobile phone or PDA that incorporates an accelerometer. We have gathered suggestions of approaches in order to build the system described in the given sections. The proposed solution can furnish better functionalities and is very easy to deploy in a regular mobile phone or PDA with an accelerometer. This application was designed to notify and monitor falls, thus providing an instant help not only in hospital but also on daily life environments. Outside the hospital, this system may be very useful for old people that live alone or people with physical diseases, for example.
9
3. System Architecture This sector discusses the requirements of the SensorFall tool, the application architecture, application development, and the technologies necessary for its deployment. Our emphasis will be on obtainable as well as emerging architectures that foster its practical deployment.
Figure 1 – Fall detection and alerting system.
The system consists of several components, accelerometer, which measures the accelerations, XML file to store the information, use of GPS to get the coordinates, and the monitoring of the accelerations. In the events we have sound, vibration and connection to a GSM service which have protocol for sending SMS messages and calls.
10
Figure 1 shows the relationships of the application with the various components that make up the application, such as Alert parameters (GPS, Call, SMS, Sound), and accelerometer, shows the interaction by accelerometer, when parameterize the device to the patient, also shows that when a fall is detected the application sends alarm to the contacts and medical help, by SMS, phone calls, GPS position, and sound issues, depending on the options that are chosen. 3.1 Application Architecture The application SensorFall is constantly monitoring the patient, using the accelerometer to measure acceleration, process the information obtained by them and provides in the display of the PDA. Data from the menus and options are saved in the XML file, and contacts use the Microsoft Outlook from the PDA to store them, and can also save a contact on the PDA and are automatically obtained by SensorFall. When application is started, the data is loaded from the XML file and will apply it to behave according to these parameters. 3.2 System Requirements A mobile application has specific requirements that SensorFall must also cope with. The user interface must be as easy to use as possible, with minimal input from the user. Screen size and orientation are a big concern in this system. Stylus input minimization dictate an interface with large buttons and appropriate font size, to enable on-the-move application use. SensorFall is not just another mobile application, it offers a host of features that are all combined in one gadget. It must offer constant monitoring and deal with several variants of accelerations. 11
To better explain the main steps of SensorFall, the corresponding activity diagram is shown in Figure 2. The activity diagram is used to describe operational activity (step-by-step) of system components, following the global flow of control. In SensorFall, when a drop is detected, the flow begins whether it is a fall or just a false positive. If it is a not fall, it goes back to monitoring, otherwise, because it said YES or because there was no response in last few seconds, it goes to the next stage in which verifies what parameters will be used, also verifying if everything was send correctly. If user selects NO, it tries again; if YES, it comes back to the monitoring. 3.3 Technologies The mobile application targets Pocket PCs running Windows Mobile 6 and 6.1 (WM). Two major solutions for application development exist: .NET Compact Framework and J2ME. The Microsoft solution was chosen, mainly due to the following two features: immediate availability of version 2.0 on every WM 6.0 device without the need for further installations, and tight interaction with a suitable database engine. The development software used to create the mobile application was the Microsoft Visual Studio 2008 PRO IDE that offers the Microsoft Windows SDK V6.0. Both versions offer emulation debug for simple initial debugging of applications. The IDE also offers a database system management, the Microsoft SQL Server 2005 Compact Edition that we used to generate and manage the database. The C# is a language oriented to objects, which was very powerful fusion of Java with C++, also because it tries to resemble Visual Basic, and was specially made from base for the Visual Studio platform, which is dedicated to the Framework .NET. 12
Figure 2 – The Activity diagram of SensorFall.
The Microsoft .NET Compact Framework is a framework that currently can only be installed on computers having a Windows operating system (OS). It includes large programming libraries, solutions for common problems, and a virtual machine that manages the programs execution written specifically for the framework. We have chosen the .NET Framework 2.0 instead of the latest 3.5, so it can be combined with another framework, the Smart Device Framework, which provides a complete scope of the core. The smart Device Framework allows developers to concentrate on the core functionality of the application. Advantages
13
are the visual power level, thus being more attractive, with more custom buttons and flashier, and also a provided tool to manage the battery. 3.4 Application Development The interface of the mobile application is simple, with large buttons for the user, and also with standard size of Visual Studio, but with appropriated size for finger use. The information provided by the user application is concise and with an appropriate size for easy reading. When the application is running, even in the background, it will always be monitoring the values of the accelerations in several areas. It will consider the values of accelerations and whenever there is a peak acceleration that exceeds the predefined threshold value of the drop, the application launches an alert, sound, bright and vibrant (depending on the choices that are selected). The application will wait a certain period of time that user may respond to this alert. The user can confirm a fall or a false alarm. After this predefined period of time, if the application does not get response or if the YES button is not pressed, the application will send a short message service (SMS) or will make a phone call (depending on the system customization). To determine the detection of falls, it has to circumvent the so-called false positives, which can range from a jump, going down/up stairs or even sitting in a chair, such as in the case of an elderly person. This is a difficult situation to evaluate because one can literally fall on the chair and thus have a significant acceleration. In order to circumvent these obstacles, the system was tested and evaluated under several situations of gender. Based on the collected results, several graphics were built in order to understand better the collected values, and thus reaching a threshold value which shall be deemed to be exceeded fall. 14
4. System Evaluation and Validation This section presents the performance evaluation and validation of SensorFall tool. 4.1 User experience and the SensorFall interface This subsection presents a general idea of the mobile application and its use in practical deployment. The SensorFall shows a simple user interface, which is easy to use by its adopters. The main window is very intuitive. Figure 3 presents the main application interface where we can see how the screen of SensorFall is presented poorly at the start of the application. The figure is numbered so it may be easier understanding the environment of the user. In (1) is displayed the threshold value of warning. If this threshold value is passed by any of the axes, it will trigger the warning of a fall. The area identified by (2) include the information collected by each one of the three axis X, Y and Z and the PDA Orientation, with colored bars. Below the three-axis information, a bar with the percentage of available battery is shown. When the values reach the available percentage of 20%, 15%, 10% and 5% it prompts the user to charge the PDA. The Menu Button is identified by (3).
15
Figure 3 – Main Application Interface.
Depending on the PDA position in relation to land, the orientation differs, because the values of the accelerations are different in different axes. When the PDA is with the guidance Face Down, in perfect conditions, one can see the axis values X = 0, Y = 0 and Z =- 9.8 m/s2. The sensor is very sensitive and it can be a fairly accurate reading of the values. If the PDA is Face Down, the X and Y axis will present the same values close to zero, but the Z axis will be positive and close to 9.8 m/s2. With Portrait orientation of the PDA, the axis that has the value of gravity is the Y with a positive value. If the PDA is in Landscape, the X-axis measures gravity with a positive value. In case of Reverse Portrait, the Y-axis shows the negative gravity value, and in Reverse Landscape the Z-axis is also negative.
16
After clicking Menu, it will appear the Login window where user can show it is authentic with his Login and corresponding Password. If any data is incorrect the user is not allowed to go to the Options and the status of a message Login Failed will appear. After making the successful login, the application shown in the screen represented by Figure 4 will appear. This user (meaning that the user is here, which contains the permission to enter the options) is facing a window, divided by TABs in order to make it more fluid to browse. The figures are numbered so it can be easier to understand the environment of the user. In (1) appears the desktop where the user chooses the options that best suit his/her use of the apparatus. Horizontal bar is pointed by (2) and it contains several TABs, in which each presents a different desktop, with different options. Thus, it enables the separation of content areas and a greater storage of application. After concluding the application customization, the Save Button (3) can be pressed for recording data in a file.
Figure 4 – Illustration of option menu to configure warning messages.
17
The first TAB, classified Data, where the user decides whether the operation, after detection of a drop, will be to send SMS, a phone call, vibration, or if it beeps, and also choose the tone of the application. The second TAB, allows the user to configure the parameters of the sensor, which set the value of warning, and the sub-menu, calibrate, it is possible to calibrate a value thanks to the test that is done to succeed in finding the best and the most suitable value for the user of the device.
Figure 5 – Illustration of calibration process.
Figure 5 shows the sub-menu of the sensor, the part of calibration, when is pressed the Start button, the alert system of fall will hold, and will be monitored to obtain the correct value for the user. The system monitors the three axes, and when is pressed the Stop button, the value is proposed and the monitoring system of fall will be active again.
18
Figure 6 – Illustration of GPS configuration and SMS tabs.
In the third TAB, Figure 6 at left, the GPS configuration, it’s possible to set the state of the GPS, to on or off, test the GPS to see the Latitude, Longitude and the number of satellites. In the fourth TAB, Figure 6, at right, the SMS configuration, the contact parameters user name and address or room number can be set (if the application is customized for hospital environment). In the fifth TAB, is the area of contact, where the user adds, modifies and removes his/her contacts. The contacts will receive alerts, in case of fall. If it contains only one contact, only this contact will receive alerts. If there are more registered contacts, all of them will receive notifications. The last TAB contains the about. When a fall is detected a window will appear, when the PDA is dropped or there is a false positive, containing bars that surround flash to draw more attention. In the center of the screen two large buttons appear, with different colors, green for
19
YES and red for NO. If none of these two options are selected, after 10 seconds, YES option is accepted by default. 4.2 Application validation This subsection discusses the scenarios performed by the system and corresponding results obtained. Authors simulated both real situations of falls and other movements to identify false positives. Note that the values of acceleration can be negative due to the position of the accelerometer, which corresponds to the orientation of the screen. For falls detection, it is valid the threshold value of 17 and -17 (as it is not possible to have a fall up). In the first scenario, where the simulation start with walking, then happens a fall into the bed, and after that, a new fall but this time to front, in the Figure 8 shows the results collected, as it may be seen, at the time 31 the simulation of a fall is presented in bed with the PDA Facing Down. After, at the time 46 the simulation of someone raising from the ground up to 52 and starting to walk, with the PDA in Portrait orientation. Finally, at the time 91 it is registered a fall in forward motion. After detailed analysis of these and other sets of results, the threshold value of 17 was defined.
20
Figure 8 – Try-axial results for the first scenario (walk, fall in bed, and fall to front).
Another important scenario is presented in Figure 9 here the simulation start with chair lift, then walking a little, to a new moment when is performance a drop to a chair, and at last a new chair lift. At the time 13 the chair lift is presented (Face Up). From the time 19, the user is walking until the time 64, when he/she performs a drop on chair. After that, he/she lifts from the chair, time 82. Finally, at the time 97 the PDA was taken from the pocket.
21
Figure 9 – Try-axial results for the second scenario (walk, drop to the chair, and chair lift).
In the third scenario, the simulation begins with downstairs and upstairs, after those has been made some walking and tries a slow fall. In Figure 10, the first time are not worthy of record because vibrations are carried out by test preparation, starting at the time 25 in which the state is stopped until at the time 45, when he/she start down stairs, turning to stop at the time 93 and then at the time 105 start upstairs and again at the time 137the state is stop, for thus to be able to more easily distinguish between movements made and with a certain interval between them, at time 153 begins to walk and at time 181 it is registered a fall in slow motion.
22
Figure 10 – Try-axial results for the third scenario (downstairs, upstairs, walk and slow fall).
The scenario starts with some walking, downstairs and walking again, then have been perform a backward fall, walking again, and a new fall, but this time, to forward. In Figure 11, the scenario starts the simulation with walk at time 9 to 29 when it will stop, and at the time 37 begins to descend stairs, at the time 65, stopped until at time 73 where start to walk until at time 89 is a simulated fall behind, at time 109 start walking, and at time 141 registered a fall forward. As it may be seen through the performed tests, when simulating a fall, the acceleration value of 17 in one of the three axes, was confirmed as the right threshold value for the system. The day-to-day activities do not pass this threshold value, so it was defined as the threshold value of alert 17.
23
Figure 11 – Try-axial results for the fourth scenario (walk, downstairs, fall to back and fall to front).
The simulation begins with walking, some running next to a sprint and at the end, walking again. In Figure 12, the fifth scenario, the simulation starts with walking at time 13 to time 46 where begins to run, up to the time 60, which is a short pause and then starts a sprint at time 70 to start walking at time 85. As shown in the chart when it runs, the values pass the pre-default value 17, which would mean a fall in the previous scenarios, this shows how important it is to calibrate the apparatus for different users, who have different activities.
24
Figure 12 – Try-axial results for the fifth scenario (walk, run, sprint and walk).
5. Conclusions and Future Work This paper presented a mobile application tool for PDAs with an accelerometer called SensorFall. The main objective of this system is to detect and notify a fall and it has been fully achieved. It includes several features, such as sound and visual alarms, the total capacity for system customization, and with an advantage of being able to adapt to different individuals, from a senior to a sportsman. Taking into account the results presented and analyzed in this paper, we can conclude that the mobile application is able to detect with sufficient accuracy the values of acceleration, thereby enabling to create a line between an invisible fall, a true and a false positive.
25
SensorFall is easy to handle, providing a very simple user interface. The mobile application also allows a low consumption of battery, even if the application is running in the background of the OS. Is incorporated and use of global positioning system (GPS), so in case of a fall outside the hospital environment, we can obtain the coordinates of the user and that will be send by SMS asking for support, providing a more effective way to a faster and a more accurate intervention.
Acknowledgments
Part of this work has been supported by the Instituto de Telecomunicações, Next Generation Networks and Applications (NetGNA) Group, Portugal, and by the Euro-NF Network of Excellence from the Seventh Framework Programme of EU.
References [1] Wee-Soon Yeoh, Jian-Kang, Wu, Isaac Pek, Yi-Han Yong, Xiang Chen, and Agustinus Borgy Waluyo, "Real-Time Tracking of Flexion Angle by Using Wearable Accelerometer Sensors", in 5th International Workshop on Wearable and Implantable Body Sensor Networks, Chinese University of Hong Kong, China, June 1-3, 2008. [2] Amit Purwar, Do Un Jeong, Wan Young Chung, "Activity Monitoring from Real-Time Triaxial Accelerometer data using Sensor Network", in International Conference on Control, Automation and Systems (ICCAS 2007), Seoul, Korea, October 17-20, 2007. [3] Pekka Kannus, et al., “Fall-induced injuries and deaths among older adults”, The journal of the American Medical Association (JAMA), Vol. 281, No. 20, May 1999, pp. 1895-1899. [4] Erica Weir and Luana Culmer, “Fall prevention in the elderly population”. Canadian Medical Association Journal, Vol. 171 (7), September 2004. [5] Mary E. Tinetti, “Preventing falls in elderly persons”. The New England Journal of Medicine. No. 348, Issue 1, January 2003, pp. 42-49.
26
[6] Pekka Kannus, Harri Sievänen, Mika Palvanen, Teppo Jarvinen, and Jari Parkkari, “Prevention of falls and consequent injuries in elderly people”. The Lancet, Vol. 366, Isue 9500, November 2005; pp. 1885-1893. [7] Zhenyu He, Zhibin Liu, Lianwen Jin, Li-Xin Zhen, and Jian-Cheng Huang, "Weightlessness Feature - A Novel Feature for Single Triaxial Accelerometer based Activity Recognition", 19th International Conference on Pattern Recognition (ICPR 2008), Tampa, FL, USA, December 8-11, 2008. [8] Kai Kunze and Paul Lukowicz, "Using acceleration signatures from everyday activities for onbody device location", 11th IEEE International Symposium on Wearable Computers (ISWC 2007), Boston, MA, USA, October 11-13, 2007. [9] Susana Palma, Hugo Silva, Hugo Gamboa, Pedro Mil-Homens, "Standing jump loft time measurement", Biosignals 2008, Madeira, Portugal, January 28-31, 2008. [10] Wee-Soon Yeoh, Isaac Pek, Yi-Han Yong, Xiang Chen, and Agustinus Borgy Waluyo, "Ambulatory Monitoring of Human Posture and Walking Speed Using Wearable Accelerometer Sensors", 30th Annual International Conference of the. IEEE Engineering in Medicine and Biology Society, Vancouver, Canada, August 20-24, 2008. [11] Youngbum Lee, Jinkwon Kim, Muntak Son, and Myoungho Lee, "Implementation of Accelerometer Sensor Module and Fall Detection Monitoring System based on Wireless Sensor Network", in 29th Annual International Conference of the IEEE EMBS, Lyon, France, August 2326, 2007. [12] Wan-Young Chung, Sachin Bhardwaj, Amit Punvar, Dae-Seok Lee and Risto Myllylae, Member, IEEE, "A Fusion Health Monitoring Using ECG and Accelerometer sensors for Elderly Persons at Home", in 29th Annual International Conference of the IEEE EMBS, Lyon, France, August 23-26, 2007. [13] SunLee Bang, Minho Kim, Sa-kwang Song, Soo-Jun Park, "Toward Real Time Detection of the Basic Living Activity in Home Using a Wearable Sensor and Smart Home Sensors", in 30th Annual International Conference of the IEEE EMBS, Vancouver, British Columbia, Canada, August 20-24, 2008. [14] Davrondzhon Gafurov, Einar Snekkenes and Patrick Bours, "Gait Authentication and Identification Using Wearable Accelerometer Sensor", 2007 IEEE Workshop on Automatic Identification Advanced Technologies, Alghero, Italy, June 7-8, 2007.
27
[15] A. M. Khan, Y. K. Lee, and T.-S. Kim, Member, IEEE, "Accelerometer Signal-based Human Activity Recognition Using Augmented Autoregressive Model Coefficients and Artificial Neural Nets", in 30th Annual International Conference of the IEEE EMBS, Vancouver, British Columbia, Canada, August 20-24, 2008. [16] V Gemignani, E Bianchini, F Faita, M Giannoni, E Pasanisi, E Picano, T Bombardini, "Assessment of Cardiologic Systole and Diastole Duration in Exercise Stress Tests with a Transcutaneous Accelerometer Sensor", Computers in Cardiology, Bologna, Italy, September 1417, 2008. [17] Nicholas Caporusso, Irene Lasorsa, Oliviero Rinaldi, Leonardo la Pietra, “A Pervasive Solution for Risk Awareness in the context of Fall Prevention”, Pervasive Health 2009, London, UK, April 1-3, 2009, pp. 1-8. [18]N. Noury, A. Fleury, P. Rumeau, A. K. Bourke, G. Ó Laighin, V. Rialle, “ Fall detection – Principles and Methods”, in 28th Annual International Conference of the IEEE EMBS Lyon, France, August 23-28, 2007. [19] Chih-Ning Huang, Chih-Yen Chiang, Jui-Sheng Chang, Yi-Chied Chou, Y-Xuan Hong, Steen J. hsu, Woei-Chyn Chu, Chia-Tai Chan, “Location-Aware Fall Detection System for Medical Care Quality Improvement”. In Third International Conference on Multimedia and Ubiquitous Engineering (MUE 2009), Qingdao, China, June 4-6, 2009, pp. 477-480. [20] Anh Diuh, Daniel Teng, Li Chen, Yang Shi, Carl McCrosky, Jenny Basran, Vanima Del Bello-Hass, “Implementation of a Physical Activity monitoring System for The Elderly People With Built-in Vital Sign and Fall Detection”. In Sixth International Conference on Information Technology: New Generations, Las Vegas, NV, USA, April 27-29, 2009, pp. 1226-1231. [21] Federico Bianchi, Stephen J. Redmond, Michael R. Narayanan, Sergio Cerutti, Branko G. Celler, Nigel H. Lovell, “Falls Event Detection using Triaxial Accelerometry and Barometric Pressure Measurement”, in 31th Annual International IEEE EMBS Conference, Minneapolis, Minnesota, USA, September 2-6, 2009. [22] Miguel Àngel Estudillo-Valderrama, Laura M. Roa, Javier reina-Tosina, David NaranjoHernández, ”Design and Implementation of a Distributed Fall Detection System – Personal Server”, IEEE Transactions on information technology in biomedicine, Vol. 13, No. 6, November 2009, pp. 874-881.
28
[23] Qiang Li, John A. Stankovic, Mark A. Hanson, Adam T. Barth, John lach, “Accurate, Fast Fall Detection using Gyroscopes and Accelerometer-Derived Posture Information”, in Sixth International Workshop on Wearable and implementable Body Sensor Networks (BSN 2009), Berkeley, CA, USA, June 3-5, 2009, pp. 138-143. [24] Xiaodan Zhuang, Jing Huang, Gerasimos Potaminanos, and Mark Hasegawa-Johnson, ”Acoustic Fall Detection using Gaussian Mixture Models and GMM Supervectors”. In IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2009), April 1924, pp. 69-72. [25] Tong Zhang, Jue Wang, Ping Liu, and Jing Hou, “Fall Detection by Embedding an Accelerometer in Cellphone and Using KFD Algorithm”. In SNS International Journal of Computer Science and Network Security, Vol. 6, No. 10, October 2006, pp. 277-284.
29
Ivo Miguel C. M. Lopes is a Master student on Informatics Engineering at the University of Beira Interior, Covilhã, Portugal, under supervision of Prof. Joel J. P. C. Rodrigues. He received his BSc degree in Informatics Engineering from University of Beira Interior, 2009. His research interests include mobile computing, eHealth and sensor networks. He is co-author of an international conference paper.
Binod Vaidya received M.S degree in Radio Communication Engineering from Odessa Electro-technical Institute of Communications, Ukraine in 1997 and Ph.D degree in Information and Communication Engineering from Chosun University, Korea in 2007. Since 1997, he have been working as a Lecturer in Institute of Engineering, Tribhuvan University, Nepal, however, from 2004 he is in academic leave. During the period of 1997 - 2004, he served various positions in Institute of Engineering, Tribhuvan University, Nepal. Dr. Vaidya was a Postdoctoral Researcher in Chosun University from September 2007 to August 2008, and a Research Associate in Gwangju Institute of Science and Technology (GIST) Korea from September 2008 to February 2009. Currently he is affiliated with Instituto de Telecomunicações, Portugal since March 2009. Dr. Vaidya has authored or co-authored over 50 papers in international journals, books, conferences, and symposia. He has served as chairs for several International conferences and workshops and also as TPC members for various conferences and workshops. Dr. Vaidya serves as Editorial Board Member for IJEHMC, IGI Publisher and JCIT, AICIT. He is also a guest editor for several International Journals such as Springer’s Telecommunication Systems and Soft Computing. His 30
current research interests are ubiquitous computing, wireless ad-hoc and sensor networks, resilience, security, and delay tolerant networks.
Joel José P. C. Rodrigues is a Professor at the Department of Informatics of the University of Beira Interior, Covilhã, Portugal,
and
researcher
at
the
Instituto
de
Telecomunicações, Portugal. He received a PhD degree in Informatics Engineering, MSc degree from the University of Beira Interior, Portugal, and a 5-year B.S. degree (licentiate) in Informatics Engineering from University of Coimbra, Portugal. His main research interests include delay-tolerant networks, sensor networks, high-speed networks, elearning, e-Health, and mo- bile and ubiquitous computing. He is the leader of NetGNA Research Group (http://netgna.it.ubi.pt). He is the Secretary of the IEEE ComSoc Technical Committee on e-health and the Vice-Chair of the IEEE ComSoc Technical Committee on Communications Software. He is the Editor-inChief of the International Journal on E-Health and Medical Communications. He participates in several European networks of excellence and Portuguese research projects. He chaired TPCs at major conferences and participated in tens of TPCs. He has been guest editor of several journal special issues and belongs to numerous editorial review boards of international journals. Joel has authored or co-authored over 150 papers in refereed international journals and conferences, book chapters, a book, and a patent. He is a licensed professional engineer and member of the ACM SIGCOMM, the Internet Society, an IARIA Fellow, and a senior member of the IEEE.
31
