The Elderly Fall Risk Assessment and Prediction

0 downloads 0 Views 1MB Size Report
Zaigham Mahmood. Towards a .... A Review of Gradient-Based and Edge-Based Feature Extraction Methods for Object. Detection. ..... Service Tailoring: Supporting Independent Living of Elderly with IT ......................................................................467 .... Frédéric Pinel, Johnatan E. Pecero, Pascal Bouvry, and Samee U. Khan.
CIT 2011 The 11th IEEE International Conference on Computer and Information Technology The 11th IEEE International Conference on Scalable Computing and Communications 31 August–2 September 2011 Pafos, Cyprus

Conference Information Papers by Session Papers by Author Search Getting Started Trademarks

Proceedings 11th IEEE International Conference on Computer and Information Technology Pafos, Cyprus 31 August–2 September 2011

Los Alamitos, California Washington



Tokyo

2011 11th IEEE International Conference on Computer and Information Technology

CIT 2011 Table of Contents Introduction................................................................................................................................................................xiv Message from the CIT 2011 General Chair..............................................................................................................xv Message from the CIT-2011 Steering Chairs...........................................................................................................xvi CIT 2011 Committees ...............................................................................................................................................xvii Message from the SCALCOM-2011 General Chairs............................................................................................xxiii Message from the SCALCOM-2011 Steering Committee....................................................................................xxiv SCALCOM 2011 Committees..................................................................................................................................xxv Message from the P2P-RDM 2011 Workshop Chairs .........................................................................................xxvii P2P-RDM 2011 Committees..................................................................................................................................xxviii Message from DSOC 2011 Workshop Chairs........................................................................................................xxix DSOC 2011 Committees............................................................................................................................................xxx Message from SCALSOL 2011 Workshop Chairs................................................................................................xxxi SCALSOL 2011 Committees..................................................................................................................................xxxii Message from the SLSIS 2011 Workshop Chairs...............................................................................................xxxiii SLSIS 2011 Committees.........................................................................................................................................xxxiv Keynote Addresses...................................................................................................................................................xxxv

SCALCOM 1: Networking and Applications Scalable Wide-Area Multicast with Temporal Rate Filtering Distribution Framework ................................................1 Harald Gjermundrod, Carl Hauser, and David Bakken The Potential of Using Network Coding with Geographical Forwarding Routing for Wireless Multimedia Sensor Networks ....................................................................................................................9 Fahed Awad, Omar Banimelhem, and Nadia Al-Rousan An Energy-Aware Clustering Scheme for Mobile Applications ..................................................................................15 Carmela Comito, Domenico Talia, and Paolo Trunfio

v

Dynamic Active Window Management: A Method for Improving Revenue Generation in Dynamic Enterprise Systems ....................................................................................................................................23 M. Al-Ghamdi, A.P. Chester, L. He, S.A. Jarvis, and J.W.J. Xue

P2P-RDM Workshop DRing: A Layered Scheme for Range Queries over DHTs ..........................................................................................29 Nicolas Hidalgo, Erika Rosas, Luciana Arantes, Olivier Marin, Pierre Sens, and Xavier Bonnaire VoroStore—A Secure and Reliable Data Storage for Peer-to-Peer-Based MMVEs ...................................................35 Sebastian Holzapfel, Sebastian Schuster, and Torben Weis Implementation and Evaluation of a P2P Service Discovery System: Application in a Dynamic Large Scale Computing Infrastructure .......................................................................................................41 Eddy Caron, Florent Chuffart, Haiwu He, and Cédric Tedeschi Probabilistic Dropping in Push and Pull Dissemination over Distributed Hash Tables ...............................................47 Emanuele Carlini, Massimo Coppola, and Laura Ricci Experiences with Complex User Profiles for Approximate P2P Community Matching .............................................53 Patrizio Dazzi, Matteo Mordacchini, and Fabio Baglini

SCALCOM 2: Parallel and Cloud Computing An Abstraction to Support Design of Deadlock-free Routing Algorithms for Large and Hierarchical NoCs ..................................................................................................................................................59 Rickard Holsmark and Shashi Kumar A Framework for Data Center Scale Dynamic Resource Allocation Algorithms ........................................................67 A.P. Chester, M. Leeke, M. Al-Ghamdi, A. Jhumka, and S.A. Jarvis A Grid-Based Cloaking Scheme for Continuous Queries in Distributed Systems .......................................................75 Hyeong-Il Kim, Young-sung Shin, and Jae-Woo Chang Proximity-Aware Resource Discovery Architecture in Peer-to-Peer Based Volunteer Computing System .......................................................................................................................................................83 Toktam Ghafarian-M., Hossein Deldari, and Mohhamad-H. Yaghmaee-M. Accelerating Circle Detection Based on Generalized Projection Method with GPUs .................................................91 Su Chen and Hai Jiang

DSOC Workshop An Efficient Role Based Access Control System for Cloud Computing .....................................................................97 Zhu Tianyi, Liu Weidong, and Song Jiaxing WSRank: A Collaborative Ranking Approach for Web Service Selection ................................................................103 Linlin Meng, Jianxin Li, and Hailong Sun

vi

SCENETester: A Testing Framework to Support Fault Diagnosis for Web Service Composition ................................................................................................................................................................109 Zekun Zhu, Jianxin Li, Yongwang Zhao, and Zhuqing Li Middleware Services at Cloud Virtual Layer .............................................................................................................115 Imad M. Abbadi Cloud Computing: Characteristics and Deployment Approaches ..............................................................................121 Zaigham Mahmood Towards a Data Complexity Metric Set for Web Service Composition ....................................................................127 Chengying Mao Critical Review of Analytical Modelling Approaches for Performability Evaluation of the Handover Phenomena in Mobile Communication Systems .............................................................................132 Y. Kirsal, E. Ever, O. Gemikonakli, and G. Mapp SOHO Network Performance Optimization via Local Caching with the Presence of a Main Bottleneck, the Accessing WAN Link ...........................................................................................................138 Daniel James Waters and Jianjun Zhang

CIT 1: Graphics and Image Processing MRF-based Particle Filters for Multi-touch Tracking and Gesture Likelihoods .......................................................144 Chi-Min Oh, Md. Zahidul Islam, and Chil-Woo Lee Geometric Active Model for Lesion Segmentation on Breast Ultrasound Images ....................................................150 Myungeun Lee, Yanjuan Chen, Soohyung Kim, and Kwanggi Kim A Bit Collision Detection Based Hybrid Query Tree Protocol for Anti-collision in RFID System ..........................................................................................................................................................158 Haosong Gou and Younghwan Yoo A Secure Recognition Based Graphical Password by Watermarking ........................................................................164 Arash Habibi Lashkari, Azizah Abdul Manaf, and Maslin Masrom Vehicle Detection on Aerial Images by Extracting Corner Features for Rotational Invariant Shape Matching ...........................................................................................................................................171 Sheng Wang The Elderly Fall Risk Assessment and Prediction Based on Gait Analysis ...............................................................176 Susu Jiang, Bofeng Zhang, and Daming Wei

CIT 2: Computer Architecture and Networks Change Function of 2D/3D Network-on-Chip ...........................................................................................................181 Alexander Wei Yin, Thomas Canhao Xu, Bo Yang, Pasi Liljeberg, and Hannu Tenhunen Designing Efficient Parallel Prefix Sum Algorithms for GPUs .................................................................................189 Gabriele Capannini

vii

Wavelet Filter Bank-Based Non Uniform Multi-tone Transceiver for Digital Subscriber Line ...........................................................................................................................................................197 Samah Mustafa, Velar Hikmat, and Salar Shekha Time Synchronization of Distributed Readers for a Large-Scale Active RFID Network ..........................................204 Hyuntae Cho, Sanghyun Son, Jongdeok Kim, and Yunju Baek Design and Implementation of MAC Protocol for SmartGrid HAN Environment ....................................................212 Min Seok Kim, Sung Ryul Kim, Jeonghyun Kim, and Younghwan Yoo Policy-Driven Reconfiguration Incorporating Multi-objective Optimization for Performance Management in a Ship Backbone Network .....................................................................................218 Sungwoo Tak and Hyejin Kim

CIT 3: Computer Networks LHCDS: A Novel Deployment Strategy of Proxy Caches for P2P Traffic in ISP Networks .....................................................................................................................................................................224 Haibin Zhai, Hai Jiang, Yi Sun, Jun Li, Jing Liu, Gengfa Fang, and Eryk Dutkiewicz A Full-Distributed Architecture for PoC Application in Data Packet Voice Communication ..........................................................................................................................................................231 Qi Wang, Hai Jiang, Albert K. Wong, Jun Li, and Zhongcheng Li Novel Multicast Operation Method in Metro Ethernet Networks ..............................................................................238 Omayma A. El-Mohsen and Hussein M. Harb Modified Deterministic Packet Marking for DDoS Attack Traceback in IPv6 Network ...........................................245 You-ye Sun, Cui Zhang, Shao-qing Meng, and Kai-ning Lu Survey, Analysis and Re-evaluation—How Efficient and Secure a Mix Network Can Be ................................................................................................................................................................................249 Kun Peng A Poisson Based Bursty Model of Internet Traffic ....................................................................................................255 Edward Kresch and Sarvesh Kulkarni

CIT 4: Graphics and Image Processing A New Human Interactive Proof System Using Arbitrary and Fractal Polygon Image .............................................261 Seon Yeong Kim, Sora Kim, and Hwan-Gue Cho Decreasing Change Impact Using Smart LSB Pixel Mapping and Data Rearrangement ..........................................269 Septimiu Fabian Mare, Mircea Vladutiu, and Lucian Prodan A Review of Gradient-Based and Edge-Based Feature Extraction Methods for Object Detection .....................................................................................................................................................................277 Sheng Wang

viii

CIT 5: Graphics and Image Processing Extracting the Path of Frame Center Points Using Spatial Transformation and Motion Estimation ...................................................................................................................................................................283 Toan Nguyen Dinh, Jeong Hwan Kim, and Gueesang Lee GPU Implementation of a Region Based Algorithm for Large Images Segmentation ..............................................291 Gilles Perrot, Stéphane Domas, Raphaël Couturier, and Nicolas Bertaux Binarization of Degraded Characters Using Tensor Voting Based Color Clustering ................................................299 Kavitha Madhubalan and Gueesang Lee Lanes Detection in PCR Gel Electrophoresis Images ................................................................................................306 Sang Cheol Park, In Seop Na, Soo Hyung Kim, Guee Sang Lee, Kang Han Oh, Jeong Hwan Kim, and Tae Ho Han Open Multi Processing (OpenMP) of Gauss-Jordan Method for Solving System of Linear Equations ....................................................................................................................................................314 Panagiotis D. Michailidis and Konstantinos G. Margaritis

CIT 6: Software Engineering Starting Model-Based Testing Based on Existing Test Cases Used for Model Creation ...........................................320 Christoph Torens, Lars Ebrecht, and Karsten Lemmer Model Tree Based Adaption Strategy for Software Effort Estimation by Analogy ...................................................328 Mohammad Azzeh A Bug Rule Based Technique with Feedback for Classifying Bug Reports ..............................................................336 Tao Zhang and Byungjeong Lee A Software Architecture for Provision of Context-Aware Web-based m-Commerce Applications .........................................................................................................................................344 Benou Poulcheria and Vassilakis Costas Why Groups are Used in Software System Modernization Decisions? Comparing Group Decision-Making in Private and Public Sector ...............................................................................................350 Miia-Maarit Saarelainen

CIT 7: Cloud Computing User-Defined Adaptive Fault-Tolerant Execution of Workflows in the Grid ............................................................356 Felipe Pontes Guimaraes and Alba Cristina Magalhaes Alves de Melo AOP4CSM: An Aspect-Oriented Programming Approach for Cloud Service Monitoring ..................................................................................................................................................................363 Afef Mdhaffar, Riadh Ben Halima, Ernst Juhnke, Mohamed Jmaiel, and Bernd Freisleben Detecting Behavioral Variations in System Resources of Large Data Centers ..........................................................371 Sara Casolari, Michele Colajanni, and Stefania Tosi

ix

Application of Live Video Streaming over GRID and Cloud Infrastructures ............................................................379 Dimitris Karakasilis, Fotis Georgatos, Lambros Lambrinos, and Theodoros Alexopoulos On Application-Level Approaches to Cloudy Computing Service in Agent-Base Distributed System .....................................................................................................................................................384 Mi-Young Kang, Hyung-Ok Lee, and Ji-Seung Nam A Software Architecture for the Analysis of Large Sets of Data Streams in Cloud Infrastructures .............................................................................................................................................................389 Mauro Andreolini, Michele Colajanni, and Stefania Tosi

CIT 8: Data Management and Visualization Using Network Science to Understand the Structure of Brazilian Popular Music .....................................................395 Charith Gunaratna and Ronaldo Menezes An Interactive Web-Based Visualization Tool in Action: User Testing and Usability Aspects ........................................................................................................................................................................403 Bahtijar Vogel, Arianit Kurti, Marcelo Milrad, and Andreas Kerren Navigating Measurements by Cross-Classifications in Multi-use Data Warehouse ..................................................409 Jie Song, Yubin Bao, and Zhiliang Zhu Design of a Hierarchical Based DHT Overlay P2P Routing Algorithm ....................................................................415 Patrik Emanuel Mezö, Mircea Vladutiu, and Lucian Prodan

CIT 9: Security and Web Applications A New K-NN Query Processing Algorithm Enhancing Privacy Protection in Location-Based Services ........................................................................................................................................421 Mi Young Jang and Jae Woo Chang Independent and Personal SMS Spam Filtering .........................................................................................................429 M. Taufiq Nuruzzaman, Changmoo Lee, and Deokjai Choi Greedy and Randomized Feature Selection for Web Search Ranking .......................................................................436 Feng Pan, Tim Converse, David Ahn, Franco Salvetti, and Gianluca Donato Generation of Semantic Interactive Environment for Personalized Search ...............................................................443 Jie Yu, Jie Gong, and Fangfang Liu Predicting the Virtual Temperature of Web-Blog Articles as a Measurement Tool for Online Popularity ..................................................................................................................................................449 Su-Do Kim, Sung-Hwan Kim, and Hwan-Gue Cho Utilizing SSTAG: A Novel Tag Recommendation Mechanism to Web Page Search ...............................................455 Guijia He, Tao Zhang, Byungjeong Lee, and Jin Suk Kim A Fully Automatic Approach for Fixing Firewall Misconfigurations ........................................................................461 Nihel Ben Youssef Ben Souayeh and Adel Bouhoula

x

CIT 10: IT and E-Health Systems Flexible Homecare Application Personalization and Integration Using Pattern-Based Service Tailoring: Supporting Independent Living of Elderly with IT ......................................................................467 Mohammad Zarifi Eslami, Alireza Zarghami, Brahmananda Sapkota, and Marten van Sinderen The Use of Electronic Health Record in Greece: Current Status ...............................................................................475 Leonidas L. Fragidis and Prodromos D. Chatzoglou Contourlet-based Feature Extraction for Computer Aided Diagnosis of Medical Patterns .......................................................................................................................................................................481 Sherin M. Youssef, Ezzat A. Korany, and Rana M. Salem Spatial Cloaking Method Based on Reciprocity Property for Users’ Privacy in Road Networks .....................................................................................................................................................................487 Amina Hossain, Al-Amin Hossain, and Jae-Woo Chang

CIT 11: Computer Networks Opportunistic Scheduling and Performance Analysis on Wireless Network Coding ................................................491 Rui Zhang, Quan Qian, Mingjun Xin, and Bofeng Zhang Extending the Lifetime of Heterogeneous Sensor Networks Using a Two-Level Topology .....................................................................................................................................................................499 Mei Wu and Martin Collier Fast Augmentation Algorithms for Maximising the Flow in Repairable Flow Networks After a Component Failure .........................................................................................................................................505 Michael T. Todinov Design and Implementation of a Smartphone-Based Reliable Real-Time Wi-Fi Broadcast System .......................................................................................................................................................513 Se-Mi Kim, Seung-Chur Yang, and Jong-Deok Kim

CIT 12: Software Engineering and Security A Novel Adjustable Matrix Bloom Filter-Based Copy Detection System for Digital Libraries ......................................................................................................................................................................518 Shahabeddin Geravand and Mahmood Ahmadi A Global Dictionary Based Approach to Fast Similar Text Search in Document Repository ...................................................................................................................................................................526 Sun-Young Park, Seon Yeong Kim, Sung-Hwan Kim, and Hwan-Gue Cho ArchMDE Approach for the Formal Verification of Real Time Systems ..................................................................533 Nourchène Elleuch, Adel Khalfallah, and Samir Ben Ahmed A Clustering Data Fusion Method for Intrusion Detection System ...........................................................................539 B.A. Fessi, S. Ben Abdallah, Y. Djemaiel, and N. Boudriga

xi

Effects of Radio Triggered Sensor MAC Protocol over Wireless Sensor Network ...................................................546 Pranesh Sthapit and Jae-Young Pyun Effective SQL Injection Attack Reconstruction Using Network Recording ..............................................................552 Allen Pomeroy and Qing Tan

CIT 13: AI and Multi-agent Systems Extending Linear Discriminant Analysis by Using Unlabeled Data ..........................................................................557 Young Tae Lee, Yong Joon Shin, and Cheong Hee Park Efficient Team Formation Based on Learning and Reorganization and Influence of Communication Delay ............................................................................................................................................563 Ryota Katayanagi and Toshiharu Sugawara Machine Learning Approach to the Power Management of Server Clusters .............................................................571 Satoru Ohta and Takehito Hirota New Approach to System Level Self-Diagnosis ........................................................................................................579 Viktor Mashkov

CIT 14: Ubiquitous Computing and Computer Architecture Design and Evaluation of Hybrid Congestion Control Mechanism for Video Streaming .........................................585 Hiroki Oda, Hiroyuki Hisamatsu, and Hiroshi Noborio Exploring the Effect of Buffer Behaviour on Perceived Video Quality .....................................................................591 Colin Bailey and Xiao-Hong Peng A Novel Mobility Model for Realistic Behavior in Vehicular Ad Hoc Network ......................................................597 Dhananjay S. Gaikwad and Mukesh Zaveri A Laxity-Aware Memory Access Scheduler for High Performance Multimedia SoC ..............................................603 Guangfei Zhang, Yifei Jiang, Wenxiang Wang, and Menghao Su Software and Hardware Co-designed Multi-level TLBs for Chip Multiprocessors ...................................................609 Xiaohui Zhang, Ming Cong, and Guangqiang Chen

SCALSOL Workshop A Review on Task Performance Prediction in Multi-core Based Systems ................................................................615 Frédéric Pinel, Johnatan E. Pecero, Pascal Bouvry, and Samee U. Khan An Instruction-Level Energy Estimation and Optimization Methodology for GPU ..................................................621 Yue Wang and Nagarajan Ranganathan Green Flexible Opportunistic Computing with Virtualization ...................................................................................629 Harold Castro, German Sotelo, Cesar O. Diaz, and Pascal Bouvry Energy Efficiency on Scalable Computing Architectures ..........................................................................................635 Carlos J. Barrios Hernández, Daniel A. Sierra, Sebastien Varrette, and Dino Lopez Pacheco

xii

Scalable and Energy-Efficient Scheduling Techniques for Large-Scale Systems .....................................................641 Cesar O. Diaz, Mateusz Guzek, Johnatan E. Pecero, Pascal Bouvry, and Samee U. Khan

SLSIS Workshop A Privacy Preserving System for Cloud Computing ..................................................................................................648 Ulrich Greveler, Benjamin Justus, and Dennis Loehr Survivable Interaction Distribution Networks ............................................................................................................654 Shun-Yun Hu Author Index .............................................................................................................................................................660

xiii

2011 11th IEEE International Conference on Computer and Information Technology

The Elderly Fall Risk Assessment and Prediction Based on Gait Analysis Susu Jiang1, Bofeng Zhang1, Daming Wei2 1

School of Computer Engineering and Science, Shanghai University, 200072, Shanghai, China 2 Department of Computer Software, the University of Aizu, Fukushima, 956-8580 Japan [email protected] for accident fall. Accident fall of the elderly people often causes serious physical injury and other disease, even death. When someone get elder, he or she may walk slower than younger of them. As early as possible to find the trend of unstable gait is very important, then to remind the elderly and their families to prevent accident fall in daily life. No single measurement tool can evaluate all aspects of fall risk factors. But it may be possible that a single tool can optimally identify people at an elevated risk for falling [2]. The Berg Balance Scale (BBS) is a valid and reliable scale for the assessment of balance in older adults. But the BBS consists of 14 balance tasks and it requires very long period (for 12 months) and very hard work (participants log follow-up) for collecting these tasks data. The British STRATIFY tool was previously developed to predict falls in hospital. Although the tool had several strengths, certain limitations exist which may not be acclimatized in other countries [3]. Also in Australia, four Fall Risk Assessment Tools have been studied and the conclusion is that the tools have poor clinical properties for identifying fall risk factors for those most at risk of falling [4]. So we here do not use institution-focused nursing assessment scales but use functional assessment measures. In this paper, we use a wearable device to collect the movement parameters of human walking gait, and propose a novel fall prediction method based on our gait analysis algorithm and a tri-axial accelerometer. Nowadays the operational simplicity and better technical capacity of accelerometer suggested the method suitable for normal environment and not limited to use in a laboratory [5]. Firstly, the gait data must be collected while the tested subject walked normally. Secondly, the raw data should be clarified and reduced some noises for getting the effective gait data. Thirdly, a status level of gait of tested subject will be generated by gait analysis and some calculation.

Abstract—In this paper, a cell phone and a three dimension accelerometer are used to develop a simple fall risk prediction system. Accelerometer offers a practical and low cost method of objectively monitoring human walking. This paper gives the gait stability and gait symmetry definition only under the data conditions of acceleration. According to the gait model, the stability and symmetry of gait are analyzed and calculated, at last not only the individual's gait can be evaluated, but it also constituted a gait assessment model. The proposed gait assessment model can be used to predict the fall risk of walking subject. The experiment results show that this method achieved better performance of prediction and efficiency. Keywords- fall risk, elderly person, fall prediction, gait symmetry, gait stability, gait analysis, human walking

I.

INTRODUCTION

According to the United Nations Population Division statistics, by the end of 2009 the elderly population has reached 737 millions, accounting for 10.8% of the total population. Then, in the year 2025 it may account for 15% of the total population [1]. So, mankind is aging. Particularly in the Asia Pacific region the elderly population has grown rapidly. In China, by the end of 2009, the number of elderly people aged over 60 increased by 4.53% over the previous year. Each region in the world, hundreds of thousands of elderly people catch a number of diseases caused by accident falling. Accident fall is one of important factors on health risk that diminishes the elderly life quality. Then, they can not take care of themselves in daily life. Also it greatly increased the cost of personal and national health expenditures. Medical research shows that aging process in humans, with the recession of nervous system function and physiological function, it makes elder person reduced the ability of walking. So the elder person may easily be injured at walking, then cause many senile diseases. Fall risk evaluation and prediction are essential while the elderly population surges. Unfortunately, existing solutions are inconvenient because most of them only collect human walking data in a short time or in laboratories. Unable to test in a special laboratory to evaluate the gait of walking for each elderly condition, doctors now mostly rely on vision and experience to examine the gait stability and symmetry in the clinical situation. This assessment is a lack of objectivity in fact. It is not easy to detect gait abnormality in forepart. And if the doctor is unable to make the correct diagnosis, the elder person will not take appropriate preventive measures 978-0-7695-4388-8/11 $26.00 © 2011 IEEE DOI 10.1109/CIT.2011.82

II.

NORMAL GAIT ANALYSIS

Usually clinical medicine only concerned with the patient's gait analysis while normal gait often be ignored for analysis. But to prevention medicine, early detection some signs of the disease is very important. Gait is the external performance of the human body structure, function, exercise conditioning system, and so on when human walk. Limbs, trunk, nerve-conditioning system and systemic diseases will affect a person's gait. Normal gait is the health of people with a sense of self, the most natural, most comfortable posture when walking gait. Healthy people walk on two legs,

176

generally able to automatically adjust the position to achieve balance and stability. Normal walking legs with alternating arms naturally swing movement, and the movement direction of one arm is different with the ipsilateral leg, but consistent with the other side of the leg. The pelvis is implicated by arm swing, resulting in periodic rotation and incline. Also ankle, knee and hip angle changes in the process of motion for coordination. So the normal gait is periodic, with coordination and balance about the characteristics [7]. The walking speed decreased with the increasing age, and the increasing age decreased the greatest walking speed more significantly than the decline of comfortable walking speed. There is a hope to have a way to judge more precisely a person's gait is normal or not. Through quantitative analysis of gait stability and gait symmetry, it has obtained a series of parameters results. On this basis and colligated other factors, it can construct an early warning device to tell whether the measured subject signs of instability when walking, or a fall hazard.

Where Ci is the data of the preceding cycle, and Ci+1 is the data of the next cycle. The dynamic symmetry of gait is defined as the discrepancy of bilateral data in gait cycle on all symmetric attributes [9]. Symmetry = Discrepancy (Ri, Li)

Where Ri is the data of the right-side and Li is the data of the left-side on one attribute. B. Gait data analysis Standard deviation is used to define the variability or randomness of the walking pattern. Less the amount of variability means better neuromuscular control and walking stability. Previous studies have extracted a set of hierarchical features from the Fourteen-Linkage Walking Model, such as walking cycle features, footprint features, and orbital features. Using an ameliorated Dynamic Time Warping algorithm, the similarity of dynamic gait data on step cycles for gait stability has been calculated. Also, the difference of bilateral gait dynamic data be calculated, and got the feature data of all symmetrical attributes. On this basis, some statistical analyses by age group have been carried out. It showed the relationship between attributes and aging trend [10]. When in test time, the architecture of the system consists of a tri-axial accelerometer sensor, a cell phone for recording gait data, and a computer with a set of software system for analysis. And next time they will be put together into a small box to construct a wearable device. The tri-axial accelerometer sensor is mounted on CASI (center front waist) of the tested persons. The data from tri-axial accelerometer were saved in the cell phone. And when the person walked for a while later (for example at least walking above 10 meters) the data were transferred to the computer for analysis. The data from accelerometer must do calibration and adjusting. The coordinate of acceleration data is not parallel the space coordinate because the accelerometer is set up obliquely, so it needs to normalize the coordinate. The aim of calibration is to transform raw data to acceleration of gravity. The standard of vibration amplitude is maintained in terms of electrical output of reference accelerometer corresponding to a known value of displacement. Acceleration signals are sampled at 90Hz using purpose-written software and saved on computer for subsequent analysis.

A. Walking model Although there is now a wealth of literature pertaining to maintenance of stability when standing, there is a relative paucity of information regarding the biomechanics and physiology of walking stability [6]. Also, normal walking has often been assumed to be symmetric. A few papers have discussed the kinematical symmetry in healthy adults using statistical and mathematic calculations, but no consensus exists on the methodology. Gait assessment should not only consider gait stability, but also consider gait symmetry. Various models are currently in development, and although there are some consentaneous areas, a unified model of human walking does not yet exist. Some preliminary studies of us about gait stability and gait symmetry had been done. The raw data was collected by a ViconMX system in a laboratory. After a multi-step preprocessing, it obtained some gait feature data from original gait data. A Fourteen-Linkage Walking Model is used to make the definition of the gait stability and symmetry. The FourteenLinkage Walking Model is a collection of 19 points, 14 segments and 12 joints, used to specify the position and the configuration of a human body, as shown in [8]. The position describes the location of a body segment or joint in space, measured in meters. Body segments are considered to be rigid bodies for the purposes of describing the motion of the body. 14 segments are composed of these points, such as head, shoulder, left and right upper-arms, left and right forearms, trunk, pelvis, left and right thighs, left and right shanks, and left and right feet. 12 joints between adjacent segments are composed of these segments, such as headtrunk, head-shoulder, shoulders, elbows, hips, knees and ankles. Human walking is a cyclical movement, so here use the similarity of the data between two adjacent cycles to assess the walking stability. Stability = Similarity (Ci, Ci+1)

(2)

Figure 1. The two coordinates system

The method of adjusting is to rotate the accelerometer in correct position. Adjusting rule is that the means of axial X and Y should be ‘0’ and Z should be ‘-1’. The coordinate of acceleration data is different from that of our walking model,

(1)

177

so the coordinate should be converted for gait data analysis. The comparison of two coordinates is shown in Figure 1 and the converting result example is shown as Figure 2. There were 48 participants for test. And their ages were from 20+ to 60+. They all walked generally so as to get normal gait data. For valid analysis, it just took out every person's middle period gait data with the whole walking period. The gait data from tri-axial accelerometer had three directions. It will use the vertical direction data to calculate the gait stability and use the sagittal direction (lateral) data to calculate the gait symmetry.

recognition community to compare two speech signals uttered at different speeds. The DTW is one of the most successful algorithms for spoken word recognition, and it is the application of dynamic programming techniques. The DTW is a technique that allows local flexibility in aligning time series, and it is superior to the ubiquitous Euclidean distance used for time series classification, clustering and indexing. The DTW algorithm here used to calculate the distance of two sequences with different length of them. The formula (3) has been used to calculate the gait stability and formula (4) to calculate the gait symmetry. The results of two index values for each test person are shown in Figure 3.

Figure 2. The converting result of acceleration data

The reason of selection these two directions data is based on our previous normal gait data analysis. When a person is walking, its body is really doing the cycle motions. Normal gait is regular and cyclical. But general the walking characteristics of the various cycles are not exactly the same. If someone walked perfectly with the same gait cycle, then this person may be a robot. So for gait stability, it just considered the similarity of all contiguous stride cycles. When human walks, the center of gravity height will change with his feet start motion from unilateral foot heel rising, toe lifting, swinging, heel landing and till stance, and then the other foot so do to complete the latter half walking cycle. A walking cycle is defined as one stride and it has twice double feet support stances in one cycle period. There are two steps in one stride. For gait symmetry, it considered the difference of bilateral body by comparing right-side step with left-side step. III.

Figure 3. The Stability and symmetry Indices based on Acceleration TABLE I. p1 0.100 p9 0.062 p17 0.257 p25 0.106 p33 0.408 p41 0.088

FALL RISK PREDICTION MODEL

For every person the stability and symmetry indices are calculated by following formulae. I sta =

n −1



i =1

DTW ( A i − A i + 1 ) ( n − 1 )

(3)

Here n is the number of valid strides, and Ai is the data series of one stride of acceleration on vertical direction. I sym =

m −1

∑ DTW j =1

( B j − B j + 1 ) ( m − 1)

p2 0.633 p10 0.181 p18 0.100 p26 0.128 p34 0.059 p42 0.079

THE GAIT STABILITY INDEX OF TEST PERSON Index of stability of tested person p3 p4 p5 p6 0.071 0.125 0.102 0.023 p11 p12 p13 p14 0.069 0.058 0.029 0.036 p19 p20 p21 p22 0.077 0.340 0.137 0.042 p27 p28 p29 p30 0.041 0.052 0.039 0.035 p35 p36 p37 p38 0.208 0.416 0.364 0.109 p43 p44 p45 p46 0.084 0.257 0.047 0.085

p7 0.044 p15 0.115 p23 0.026 p31 0.035 p39 0.047 p47 0.033

p8 0.193 p16 0.071 p24 0.311 p32 0.084 p40 0.056 p48 0.136

The smaller the value of indices indicated the better the situation of gait. If the formula (3) gave out larger value, it explains one’s gait was not stable enough. If the formula (4) gave out larger value, it indicates a worse gait symmetry. The indices results can see Table Ⅰand TableⅡ. The signs from p1 to p48 are tested person’s number.

(4)

Here m is the number of valid steps, and Bj is the data series of one step of acceleration on sagittal direction. The DTW of above formula refers to Dynamic Time Warping algorithm. The DTW was developed in the speech

178

TABLE II.

Index of symmetry of every tested person p2 p3 p4 p5 p6 p7 2.675 0.315 1.436 0.515 0.171 0.204 p10 p11 p12 p13 p14 p15 0.519 1.043 1.021 1.347 0.539 0.795 p18 p19 p20 p21 p22 p23 0.569 0.397 0.662 0.782 0.293 0.937 p26 p27 p28 p29 p30 p31 0.517 0.225 0.740 0.781 0.455 0.891 p34 p35 p36 p37 p38 p39 0.406 0.466 1.196 0.733 0.806 0.364 p42 p43 p44 p45 p46 p47 0.679 0.496 0.505 0.635 0.533 0.473

p1 0.850 p9 0.302 p17 1.267 p25 0.424 p33 0.479 p41 0.620

IV.

THE GAIT SYMMETRY INDEX OF TEST PERSON

Here is not intended to do fall detection, it is because there are already many examples of research and application [11]. The goal of this paper is to find the early signs of abnormal gait of normal person. These signs can be seen as warning messages of some diseases and give the alarm to tested subject. Also it may remind clinical make some prevention and remedy for some case. Here we use the clustering algorithm APCLUSTER to carry on the grouping for the tested subjects [12]. The inputs of the clustering algorithm are two vectors, one is the tested subject's gait data and another is the corresponding age data. Before the data entry the clustering algorithm for computing, they needed to be normalized, so that the two vectors may in the same range, and then the output graphics could be easier to see clearly with the clustering results. The standard grouping will be produced in the sample database. The standard gait status will be decided by clustering and to obtain meaningful results. The examples are shown in Figure 4 and Figure 5. The clustering is done on balance attributes and age data. The balance attributes include symmetry gait data and stability gait data. A greater value of vertical axis will indicate the poorer gait level.

p8 0.875 p16 0.617 p24 0.279 p32 0.380 p40 0.352 p48 0.738

The statistical method is used to get the measuring criterion, like Table Ш. Based on the calculated data, the gait status can be classified as three levels. (1) Normal: the subject walks normally, no fall risk. (2) Attentive: the subject needs care when walking. It has no serious danger, but must take care. (3) Dangerous: the subject should pay attention, or must be careful. It has fall-risk when walking. TABLE III.

GAIT STATUS MEASURING CRITERION

Criterion

DISCUSSION OF GAIT ASSESSMENT

Clustering on Z Direction Vertical ← P2

1 0.9

Description

0.8

C3 Remark

0.7

If (mean+std < Index ≤ k*mean) then the gait status is Attentive. If (Index > k*mean) then the gait status is Dangerous. Here mean is the average value of Index of all person’s, and std is the standard deviation of Index. Obviously k is the coefficient value of 3 or others.

Normalized

C2

If (Index ≤ mean+std) then the gait status is Normal. STAB on ACC

C1

← P36 ← P33

0.6

← P37 ← P20

0.5

← P24

0.4

← P17

0.3

← P44

0.2

The gait status C2 or C3 indicated individuals who should receive further comprehensive physical inspection to find the reasons and treatment to prevent falls. How to choose coefficient k is a problem. Through the analysis of statistical data, we decided to select 3. That is using 300% of the average value as the partition boundaries for the C2 and C3.

← P4 ← P15 P1P5 ←←

0.1 0

← P35

← P8 ← P10

P19 P16 P3 ←← ← P11 ← P12 P7 ← P14 ←← ← P6P13

0

0.1

← P21 ← P26 ← P18← P25 ← P9 ← P23

0.2

← P48 ← P38 P41 P46 ← P43 ← P32 ←← P42 ← ← P34 ← P28 P39 ← P22 ← P27← ←← ← P29 P31 P30

0.3

0.4 0.5 0.6 cluster number = 6

← P40 P45 ←

0.7

0.8

← P47

0.9

1

Figure 4. The clustering example of acceleration on stability with age Clustering on Y Direction Lateral ← P2

1 0.9

TABLE IV.

ESTABLISHED THE CRITERION 0.8

TABLE V.

0.7

2*mean

3*mean

0.258 1.346

0.387 2.019

Normalized

Stability Symmetry

mean + std 0.256 1.093

C1

C2(mean+std﹤ I ≤

(I ≤ mean+std)

3*mean)

0.5

← P4 ← P13 ← P17

← P36

0.4 0.3

C3(I > 3*mean)

42

5 (p17, p20, p24, p37,

Symmetry

43

4 (p4, p13, p17, p36)

1 (p2)

Converge

(omit)

p17, p36

p2

← P16 P14 ←← P5 P10

0.1

← P19 ← P3

0

1 (p2, p33, p36)

← P23

P1P8 ←← ← P15

0.2

Stability

p44)

0.6

← P12 ← P11

DELIMITED BY CRITERION Delimited by Criterion

Criterion

SYMM on ACC

Indices

Established the Criterion

0

← P7 ← P6

0.1

← P21

← P29 ← P28 ← P37 ← P20

← P18 ← P26 ← P25 P9P24 ←←

0.2

0.3

← P22

← P31 ← P38

← P33←← P35 P30 ← P34 ← P32 ← P39

← P42 ← P45 ← P41 ← P46 ← P43 ← P44

← P48

← P47

← P40

← P27

0.4 0.5 0.6 cluster number = 6

0.7

0.8

0.9

1

Figure 5. The clustering example of acceleration on symmetry with age

To put a new tested subject into the gait database, and

179

REFERENCES

then do clustering to see which group it belongs. This can get the gait status level of the tested person. Although now we just get the gait data from a tri-axial accelerometer sensor, we think we also can do so. We put a new tested person’s gait data into the database, then do clustering. So you can see he belongs to which group. To use the criteria in practice, the indices can be used separately or combined the stability index and symmetry index. For examples, when the two indicators all exceed the limits, it can determine the gait status level of a tested person. According to the criteria, the results of the test data were very clear. That is, the tested person p2 may into Dangerous level and the tested person p36 into Attentive level. This can see on Table Ⅴ. V.

[1]

United Nations,Department of Economic and Social Affairs, Population Division, “World Population Ageing 2009,” pp. 66-71 [2] Susan W. Muir, Katherine Berg, Bert Chesworth, Neil Klar, Mark Speechley, “Balance Impairment as a Risk Factor for Falls in Community-Dwelling Older Adults Who Are High Functioning: A Prospective Study”, PTJournal (2010) vol 90(3): 338-347 [3] Alexandra Papaioannou, et al,"Prediction of falls using a risk assessment tool in the acute care setting",BMC Medicine 2004 vol 2:1, doi:10.1186/1741-7015-2-1 [4] Anna L. Barker, Jennifer C. Nitz, Nancy L. Low Choy, and Terry Haines, “Measuring Fall Risk and Predicting Who Will Fall: Clinimetric Properties of Four Fall Risk Assessment Tools for Residential Aged Care”, J Gerontol A Biol Sci Med Sci (2009) 64A(8): 916-924 [5] Moe-Nilssen R, Helbostad JL, “Estimation of gait cycle characteristics by trunk accelerometry” , Journal of Biomechanics, Vol 37, No.1, pp. 121-126, JAN 2004 [6] Yasuaki OHTAKI, Muhammad ARIF, Akihiro SUZUKI, Kazuki FUJITA, Hikaru INOOKA, Ryoichi NAGATOMI and Ichiro TSUJI, “Assement of Walking Stability of Elderly by Means of Nonlinear Time-Series Analysis and Simple Accelerometry”. JSME International Journal, Series C, Vol.43, No.4(2005), pp.607-612 [7] Tsuruoka, Y. and Tsuruoka, M., "Walking stability using portable acceleration measurement system", Proc. of IEEE International Conference on Systems, Man and Cybernetics (SMC) 2009, pp.504509, 11-14 Oct. 2009 [8] Bofeng Zhang, Kanno, T., Wenxi Chen, Gengfeng Wu, Darning Wei, “Walking Stability by Age-A Feature Analysis Based on a FourteenLinkage Model,” CIT 2007. 7th IEEE International Conference on Computer and Information Technology, 2007, Page(s): 145 – 150 [9] Susu Jiang, Bofeng Zhang, Ke Yan, Daming Wei, “Primary Exploration on the Symmetry of Human Walking,” CIT '09. Ninth IEEE International Conference on Computer and Information Technology, 2009, Volume: 1, Page(s): 263 – 268 [10] Bofeng Zhang, Ke Yan, Susu Jiang, Daming Wei, “Walking stability analysis by age based on Dynamic Time Warping,” CIT 2008. 8th IEEE International Conference on Computer and Information Technology, 2008, Page(s): 544 – 548 [11] Q. Li, J. Stankovic, M. Hanson, A. Barth, J. Lach, and G. Zhou, “Accurate, Fast Fall Detection Using Gyroscopes and AccelerometerDerived Posture Information,” Body Sensor Networks, June 2009. [12] Brendan J. Frey and Delbert Dueck, “Clustering by passing messages between data points”, Science, Vol 315, No 5814, pp.972-976, February 2007

CONCLUSIONS AND FURTHER WORK

Firstly, the accelerometer offers a practical and low cost method of objectively monitoring human walking. Secondly, the paper has defined a model to predict the fall risk of walking subject. Thirdly, the paper proposed a gait assessment method to find the walking status of tested subject. It shows that the method is very simple, and there is a strong practical. Accidental falls mainly occur in the elderly and stroke patients, so the next work will be to collect gait data which contain abnormal gait and to assay the wearable device. It is a meaning work if we can monitor the irregular walking patterns, stumbles or slip, and make early warning of the possible falls in time. We may improve the analysis algorithms, and simplify the data processing or calculation. And more, it is important to automatically distinguish the action types of the person with wearable device. If it can indicate the level of fall risk gait by using a wearable device, it may be realized the real-time evaluation of walking gait and prediction of early fall risk. ACKNOWLEDGMENT This work was sponsored by Shanghai Pujiang Program and Shanghai Leading Academic Discipline Project, Project Number: J50103.

180