UBICOMM 2010 : The Fourth International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies
`
`m-Physio: Personalized Accelerometer-based Physical Rehabilitation Platform
`
`Iván Raso, Ramón Hervás, José Bravo
`Technologies and Information Systems Department
` Castilla - La Mancha University
`13071 Ciudad Real, Spain
`Email: ivanrasodiazguerra@gmail.com; ramon.hlucas@uclm.es; jose.bravo@uclm.es
`
`
`
`
`Abstract— This paper proposes a rehabilitation system based
`on new technological tendencies in the mobility and ubiquitous
`computing areas. Specialists and patients related
`to
`rehabilitation area can use this proposed system to improve the
`fulfillment of exercises and the supervision of rehabilitation
`tasks. An important current problem is that sometimes these
`activities cannot be performed efficiently due to the lack of
`time or the large distances between patient homes and
`rehabilitation centers. We have developed our system using a
`mobile device and a bracelet to capture patient’s rehabilitation
`relevant data. As a pre-process procedure, raw data output by
`mobile device accelerometer is filtered, and then we use the
`technique called Dynamic Time Warping to train and
`recognize movements. Based on this recognition, patients can
`perform rehabilitation without the continuous specialist’s
`surveillance and can be sure of its accuracy. Experimental
`results show us that our system is able to adapt itself
`dynamically to the peculiarities of each user and enhance
`healthy rehabilitation in a proactive way.
`
`Keywords- ubiquitous computing; accelerometry; physical-
`rehabilitation; mobility;
`
`I.
` INTRODUCTION
`The ubiquity of mobile devices has led to the emergence
`of personalized and adaptive services that are able to respond
`particular needs of each specific user. These services allow
`us to develop a wide range of proactive applications such as
`ambient assisted living services (e.g., assistance to elderly
`people
`[15] and chronic diseases assistance
`[22]),
`entertainment (e.g., mobile quiz games [16]), and smart
`homes
`(e.g., personalized home control
`[17] and
`visualization services [23]).
`A principal characteristic to take into account in our work
`is the capability of monitoring user movements. Motion
`recognition is a discipline that has been around us for years
`in the scientific community. Some of the related works
`address issues such as handwriting recognition, recognition
`of hand gestures, and monitoring of the user activities. Some
`of these researches have in common with our work the use of
`one
`particular
`technology:
`the
`accelerometry.
`Accelerometers are being used in many sectors and, due to
`the fast development in sensor technology, it is possible the
`integration of these sensors into every day devices [1], for
`example, into mobile devices.
`Focusing on the rehabilitation area, patients usually have
`to move about their rehabilitation center several times, but
`sometimes, factors such as lack of time and large distances
`
`
`
`their specialists, and
`to
`the number of visits
`affect
`consequently affect in the quality of the rehabilitation.
`Moreover, some patients suffer a slight incapacitate and have
`to perform part of their rehabilitation at home and they also
`need medical examination to check their evolution. Also, the
`well-known health care systems overcharge can be lesser by
`means of this kind of m-Health systems.
`The main goal of this paper is the development of a novel
`system that helps the kinds of patient mentioned above
`whenever realize their rehabilitation. Besides, physical
`rehabilitation specialists can improve the monitoring and
`supervision of tasks by using our web-based system in their
`rehabilitation center. Our whole system (web-based and
`mobile applications) lets specialists pay a better attention to
`patients and reduce the problem of performing rehabilitation
`without the attentive specialist’s eye.
`In this paper, we employ the iPhone, one of the first
`mobile devices equipped with an accelerometer. Later,
`several mobile devices such as RIM Blackberry Storm,
`Nokia N95, and Sony Ericsson W910 were equipped with
`this kind of sensor. They basically use accelerometers for
`user interaction with games [3]. Few relevant applications
`employ accelerometry with other purposes. For example,
`Sony Ericsson’s shake control allows the changing of songs
`by shaking the mobile device. However, this paper presents a
`novel application that introduces m-Health area into a new
`challenge that has not been deeply explored, the mobile-
`based rehabilitation.
`This paper is organized in five sections: Section 2
`presents some related works that apply accelerometry to the
`rehabilitation area. Section 3 presents and explains the
`proposed applications. Section 4 presents evaluation results,
`and Section 5 discusses about the future works and the
`conclusions.
`
`II. RELATED WORK
`Accelerometry has become a powerful choice for
`evaluating variability of person’s movement using these
`kinds of sensor that provide a non-invasive method of
`measurement and have a successful accuracy [4].
`The entertainment sector is one of the most influenced by
`this technology as we can see in the Nintendo Wii game
`console that uses Wii Remote and Nunchuk to control
`avatars in games by means of natural gestures, and the
`PlayStation3 with its SixAxis and DualShock 3 controllers.
`Other sectors such as motor industry use these sensors to
`control ABS systems, airbags, and for checking the correct
`
`Copyright (c) IARIA, 2010 ISBN: 978-1-61208-100-7
`
`416
`
`Petitioner Samsung Ex-1048, 0001
`
`

`

`~
`
`.f
`
`' ~~
`
`~
`
`-~ ~ Iii
`"'~~
`L
`
`(a)
`
`(b)
`
`I r
`
`S(At) = { A,,
`A, *et + S(A(t-LJ)*(l.0 - et)),
`
`if t = o
`ifO<t::;T
`
`UBICOMM 2010 : The Fourth International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies
`
`industry also uses
`working of a machine. Transport
`accelerometry to check wherever the merchandises suffer
`misshapes and their condition or their integrity has been
`damage by it [2].
`Several works [5][6][7][8][10] have shown the accuracy
`the
`accelerometer-based
`of
`physical
`rehabilitation
`monitoring such as limb’s motion, gait analysis before
`strokes, and other illness or accidents that cause malfunction
`in the physical condition of a patient.
`
`
`
`Figure 1. Application parameters (a) and stadistics gathered from the
`rehabilitation process (b).
`
`Some related works to this paper are Wiihab [20],
`Telefonica’s Rehabitic [9], and the Arteaga et al. proposal
`[19]. These projects use the accelerometers to help in the
`rehabilitation process. The Wiihab is used as an interaction
`object that encourages people to move their limbs. However,
`Rehabitic is made up of several accelerometers and a central
`device that saves the data received from the sensors, all
`complemented by a web page that the specialist uses to
`control the patients. This system gives important information
`relevant to help the patient in the rehabilitation exercises and
`contribute to the specialist’s decisions about the evolution of
`the patient. Arteaga et al. propose a set of monitoring
`devices, each of which comprises of an accelerometer and a
`beeper, LED light and vibrator to provide redundant modes
`of inappropriate posture warnings that would hopefully
`trigger self-correction.
`Other related contributions are the O’Donovan et al. [7]
`and Choquette et al. [18] proposals that allow scientific
`monitoring limb’s motion and measures of heart rate using
`Body Area Network (BAN). Nowadays, the BAN devices
`use wireless technology and have become into Wireless Area
`Network (WAN) [5]. We have found many interesting
`similarities between
`these systems and our approach.
`However, the invasive characteristic of these systems
`prejudices their common use due to its numbers of devices
`and the tedious task of getting dressed with them. At this
`
`point, our application contributes to the rehabilitation area
`with a less invasive system than the above-described projects
`do. Moreover, a common problem with these proposals is the
`high effort needed to deploy the systems. Thus, our solutions
`achieve
`the objective of ubiquitous
`rehabilitation
`performance and monitoring that enhance the accuracy, less
`invasively and reducing infrastructural needs.
`III. SYSTEM OVERVIEW
`The mobile application (Figure 1) has been developed for
`iPhone 2G devices. This device includes an LIS302DL
`MEMS smart digital accelerometer [21]. It has 3-axis
`(X,Y,Z) and includes dynamically user selectable full scales
`from ± 2g to ± 8g.
`According to some related studies [6][7][5], one of the
`best options to wear the accelerometers to the patient is a
`wearable system. The examples mentioned above were ruled
`out because their tedious wear system. We decided to use a
`bracelet that people use together with the iPhone for jogging
`or fitness.
`Before presenting the principal system’s components, we
`define what kind of rehabilitation exercises are related to this
`paper and their particular characteristics:
`• Exercises end in the same point that they start.
`• When a patient performs the exercise, it always
`starts at the same point approximately.
`• The motion of the exercise will be slow due to the
`fact that the patient is doing rehabilitation.
`Additionally, an exercise can be classified in four types:
`• Correct exercise: The patient performs the exercise
`according to the pattern generated in the training
`process and imposed by the specialist.
`• Wrong exercise: The patient performs the exercise
`according to the time limits but it was not the
`expected exercise according to the stored pattern.
`• Exercise exceeds the maximum time: The patient
`performs an exercise but out of the maximum time
`allowed.
`• Exercise does not exceed the minimum time: The
`patient completes an exercise but does not pass the
`minimum time imposed by the specialist.
`A. Filtering
`It is necessary to use a filter because the raw data of the
`accelerometer is noise and redundant. Consequently, we
`have chosen the following smoothing function for each axis
`(Equation 1):
`
`(1)
`
`S(At) is the filtered acceleration vector output and At is
`the acceleration raw vector output, which is acquired by the
`interaction device at time t. Besides α is a smoothing factor
`in the range from 0 to 1. The α factor is critical for acquiring
`valid data to be analyzed in the pattern recognition process.
`
`Copyright (c) IARIA, 2010 ISBN: 978-1-61208-100-7
`
`417
`
`Petitioner Samsung Ex-1048, 0002
`
`

`

`~:§ ~:~ ~:~
`~-•
`~-•
`a
`·~
`·~
`
`· 1
`
`•1,S
`
`-l
`
`lb)
`
`(c)
`
`Ads
`
`- Y
`- x
`
`smooth1n1 factors
`(e)g;0,1
`(b)a::0,2
`(c:)«~.3
`(d)Cl=0.4
`(e)a:O,S
`
`·1
`
`·1,S
`
`•l
`
`-1
`
`-1.S
`
`(a)
`
`·1
`
`•1,S
`
`·l
`
`·1
`
`•1.S
`
`·2
`
`Id)
`
`-2
`
`l•l
`
`CAPTURING
`
`Setting options
`
`Touch screen when ready
`
`The device save the move
`
`Store it momentanly
`
`01W
`
`NO
`
`Select the maxlmun OTW
`
`Save the training
`
`UBICOMM 2010 : The Fourth International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies
`
`We have performed several experiments to select an α
`factor for rehabilitation exercises or other movements with
`similar characteristics; different kinds of movement may
`need additional studies to select a valid α factor. Figure 2
`the
`iPhones’s
`shows different graphs captured by
`accelerometer while a patient is performing a rehabilitation
`exercise. Each graph represents the same exercise with
`different α factors. The most representative capture of the
`rehabilitation exercise was the option (a), with α factor 0.1,
`because it filters peaks (remarked with circles) that not
`contribute to define the rehabilitation exercise. In the set of
`exercises performed for this paper we chose the option (a).
`On the other hand, fewer values of the α factor are not
`characteristic
`to
`the movement
`represented by
`the
`accelerometer axes.
`
`Figure 2. Test of different smoothing factors.
`
`
`
`B. Pattern Recognition
`is a kind of pattern
`The recognition of motion
`recognition. This recognition can be performed in several
`ways such as brute force, fuzzy logic, Gabor wavelet
`transform, hidden Markov model, support vector machine,
`and neural networks [3].
`Instead of these examples, we have decided to base our
`development in the dynamic time warping (DTW) algorithm
`because it requires a simple training and its effective has
`been proved in many researches [11], [8]. Besides, it has
`been used for writing recognition [12] with great results, as
`well as for speech recognition [13].
`DTW computes the distance between two exercises A
`and B by finding the minimum path that will be represented
`with a numerical value. In our application, the averaged
`Euclidean distance defines the cost between two different
`points Ai and Bj from the rehabilitation exercise.
`C. Segmentation
`The segmentation mechanism is used to determinate the
`beginning and end of an exercise. Some related works use
`the segmentation approach [3][1]. In our case,
`the
`segmentation is necessary because the patient has to know if
`he/she is beginning the exercises in the correct position as
`well as the device has to know when the exercise begins and
`ends. In order to achieve the segmentation mechanism,
`authors such as Schlömer [14] forces users to touch a button
`for detecting the beginning and end. Other authors [3] use
`mathematical equations such as the equation (2). According
`
`Copyright (c) IARIA, 2010 ISBN: 978-1-61208-100-7
`
`to the author when this equation is bigger than 0.3 the
`exercise starts and ends when drops to bellow 0.1.
`
`(2)
`
`The first method is not directly applicable in our system;
`patients cannot touch the mobile each time they realize a
`rehabilitation exercise because it may distort the pattern
`recognition process. On the other hand, the equation (3) is
`more interesting but neither applicable to our purposes; we
`manage several consecutive exercises and it requires that
`patients cannot stop for a while until the accelerometer drops
`to the value 0.1.
`Our segmentation method is partially based on the push-
`button approach mentioned above and follows these steps:
`• Patients have to touch the mobile’s screen when they
`are ready to start the exercise.
`• Once the patient touches the screen, the mobile
`device starts a countdown (five seconds) to allow the
`patients gets ready. For example, it could be possible
`that the patients have to make a rehabilitation
`exercise with their legs and then they need time for
`returning to the start position.
`• After the mobile device countdown, it starts to
`calculate the exercise beginning and end, taking
`enough samples to represent these facts. The number
`of chosen samples was 30 after a wide testing
`process. In the rehabilitation and training process the
`segmentation is different from the capture steps
`because it is unnecessarily taken this number of
`samples again.
`• As soon as patients finish the exercise, the mobile
`device uses the last sample to recognize the end of
`the rehabilitation exercise.
`
`Figure 3. Flow charts representing the steps to capture and train an
`exercise.
`
`
`
`418
`
`Petitioner Samsung Ex-1048, 0003
`
`

`

`1. Patient Registration
`
`2. Capture and Training
`
`m-Physio
`
`4. Specialist Monitoring
`
`+I
`
`UBICOMM 2010 : The Fourth International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies
`
`This method has not only been designed to control the
`beginning and the end of exercises, but also to validate the
`device’s position whenever patients are performing
`rehabilitation at home. If the position of the mobile device is
`detected as wrong, the mobile device does not start the
`countdown and notifies the warning to the patient. In this
`case, he/she has to wear the device again. Otherwise it could
`be fatal to the rehabilitation. This step is only performed in
`the rehabilitation and training process.
`
`
`Figure 4. Flow chart that represents the steps to perform the rehabilitation
`process.
`
`D. Rehabilitation Steps
`Three important modules compound our application:
`exercise
`capture,
`exercise
`training
`and
`personal
`rehabilitation. Each module depends on the preceding
`results. The relevant modules’ steps in response to the
`interaction of the patients are explained in the flow charts
`shown in Figure 3 and Figure 4.
`• Exercise capture: Before using this module, the
`patient has to wear the mobile device and the
`specialist explains her/him
`the steps of
`the
`rehabilitation exercise. The specialist can set the
`movement’s minimum time, the accelerometer’s
`frequency, the movement’s name and the smoothing
`factor α. The smoothing factor recommended after
`our set of test is 0.1 and the frequency is 80Hz. The
`capture of the exercise give to the mobile device the
`main configuration of the movement that is used in
`the following modules. Besides, once the exercise is
`captured, the mobile device completes the necessary
`information to enable the next steps of the system
`and stored the pattern that represents the particular
`rehabilitation exercise. This information includes the
`move’s maximum time and the accelerometer’s data
`corresponded
`to
`the exercise. This process
`is
`described in Figure 3 (left)
`
`• Exercise training: Once the exercise is stored in the
`mobile database, it is necessary to train the exercise
`for being recognized when the patient begins to
`perform the rehabilitation. Whenever patients are
`performing the training, the mobile device acquires
`all the exercises performed by them and applies the
`DTW algorithm to analyze the movements. This part
`has to be performed under the supervision of the
`specialist. Depending on the specialist criteria, the
`training can be adapted to the patient needs. The
`specialist can suggest the patient not be accurate in
`the motion or, on the other hand, the specialist can
`force patients to perform more precise exercises. In
`more detail, if the training was hard because the
`injury was important, the rehabilitation will need an
`accurate exercise, otherwise if the injury was less
`relevant, the training will be leak. As we mentioned
`before, these decision belong to the specialist
`criteria. This process is detailed in Figure 3 (right)
`• Personal rehabilitation: The personal rehabilitation
`is the most important step of our system. This
`module captures the exercises that patients perform
`in their rehabilitation process and classify them. This
`process is presented in Figure 4. There are four kinds
`of output to a patient’s exercise and were defined
`previously. Once the rehabilitation ends, the mobile
`phone stores all the outputs and analyzes them to
`allow
`the
`specialist
`controls
`the patient’s
`rehabilitation. Additionally,
`the mobile device
`synchronizes all the information with a centered
`database. This information can be accessed via the
`web application.
`• Web application: The incorporation of the web
`application
`to
`the
`system complements
`the
`supervising cycle, giving to the specialists an
`efficient method to follow the patient’s evolution.
`
`
`
`Figure 5. Principal steps on mPhysio rehabilitation process.
`
`We now show the mainly steps for using our application
`to perform the rehabilitation process at home. First, a patient
`have to move about the rehabilitation center and the
`specialist studies the patient’s case to decide the suitableness
`of our system in the specific patient’s rehabilitation. Then,
`
`Copyright (c) IARIA, 2010 ISBN: 978-1-61208-100-7
`
`419
`
`Petitioner Samsung Ex-1048, 0004
`
`

`

`Codman's rehabilitation exercise
`
`Beginning and end point
`
`Medium point
`
`Gathered pattern
`
`Generic leg rehabilitation exercise
`_,-- - - - - - ,
`
`Beginning and end point
`
`Medium point
`
`T,
`
`....
`
`Gathered pattern
`
`T,
`
`30
`
`25
`
`Cl)
`C
`0 20
`·a
`.,
`·a
`.,
`"' 15
`... .,
`Cl) 10
`·c::; ...
`.,
`><
`Ul
`
`5
`
`- -Correct
`
`Wrong
`
`- -Minimum time
`
`- -Maximum time
`
`0
`
`1
`
`3
`
`5
`
`7
`
`9
`
`11 13 15 day
`
`UBICOMM 2010 : The Fourth International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies
`
`the specialist registers the patient’s personal data into the
`mobile device. Once all the needed information is stored in
`the mobile device, the specialist gaits the patient with the
`capturing and training system’s steps. When the specialist
`decides the training is enough, the patient can comeback to
`his/her home and performs the personal rehabilitation
`process. Whenever the patient ends his/her rehabilitation
`session at home, all the data is stored in the mobile device
`and is sent to the centered database through web services.
`Finally,
`the
`specialist can
`supervise
`the patient’s
`rehabilitation evolution by means of the developed web
`application. If the specialist thinks the patient is performing
`significant errors, he/she could call or send a message to the
`patient advising him/her, and if required, set up an
`appointment in the rehabilitation center. All these steps are
`summarized in Figure 5.
`IV. RESULTS AND EVALUATION
`In order to analyze the patient’s experience with the
`application and its accuracy, we have tested it with five
`patients and two different rehabilitation exercises. The
`population includes one child, two teenagers, and two adults
`(with the ages of 43 and 64). The teenager users were
`familiar with the technology, while the other users were not
`familiar with this kind of system and device.
`
`TABLE I.
`REHABILITATION AVERAGE BASED ON THE TWO
`EVALUATED EXERCISES AND FIVE PATIENTS
`Correct
`Wrong
`Min Time
`Max Time
`23.33 %
`33.33 %
`26.67 %
`16.67 %
`30.00 %
`36.67 %
`20.00 %
`13.44 %
`36.67 %
`33.33 %
`20.00 %
`10.00 %
`36.67 %
`33.33 %
`20.00 %
`10.00 %
`33.33 %
`33.33 %
`20.00 %
`13.33 %
`36.67 %
`33.33 %
`16.67 %
`13.33 %
`40.00 %
`30.00 %
`16.67 %
`13.33 %
`46.67 %
`26.67 %
`13.33 %
`13.33 %
`50.00 %
`26.67 %
`13.33 %
`10.00 %
`56.67 %
`26.67 %
`10.00 %
`6.67 %
`60.00 %
`23.33 %
`10.00 %
`6.67 %
`73.33 %
`20.00 %
`0.00 %
`6.67 %
`76.67 %
`20.00 %
`0.00 %
`3.33 %
`83.33 %
`16.67 %
`0.00 %
`0.00 %
`93.33 %
`6.67 %
`0.00 %
`0.00 %
`
`Day
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`
`The tested exercise were a shoulder and leg movement
`shown in Figure 6. The exercises were repeated 30 times
`along 15 days, which provided 450 examples for each
`exercises and each patient. The training range was from 10 to
`20 repetitions. The results in Table 1 and Figure 7 present the
`evolution of the patient’s rehabilitation. The first days, only
`one out of every four rehabilitation exercises were performed
`correctly. Without using m-Physio or another monitoring
`system, patients are not aware of the incorrect development
`of the rehabilitation process neither the specialist. This fact
`brings as consequence an inadequate physical recovery and,
`in some cases, it may worsen the injury. Our system guides
`patients since the first day of rehabilitation and enables the
`
`enhancement of the performed exercises. Moreover, the
`specialist can supervise this process and take part whenever
`necessary.
`
`
`Figure 6. Shoulder and leg rehabilitation exercises.
`
`Focusing again on tested exercises, these results show a
`high accuracy rate of 76.67% when users were using the
`application along 13 days and it improves during the next
`days rising up to 93% in the 15th day.
`
`
`
`Figure 7. Daily evolution of the patients’ rehabilitation during the tests.
`
`
`
`V. CONCLUSION
`In this paper, we have presented and evaluated a mobile-
`based rehabilitation system that can be used in rehabilitation
`centers for improving control and supervision. The proposed
`system is completely customizable, so the specialist can
`choose the position of the device, the frequency, minimum
`and maximum time of the rehabilitation exercises and the
`accuracy of the patient when they are performing the
`rehabilitation at home or without the continuous specialist’s
`surveillance at rehabilitation center. Since the applied pattern
`recognition and segmentation techniques have been proposed
`and studied previously, we have analyzed their practical
`application to physical rehabilitation and we have optimized
`these techniques to this kind of movements.
`One of the future works of our system includes the
`improvement of the segmentation process. A new technology
`implanted in the new mobile devices can be particularly
`helpful for the recognition and validation of the exercise’s
`
`Copyright (c) IARIA, 2010 ISBN: 978-1-61208-100-7
`
`420
`
`Petitioner Samsung Ex-1048, 0005
`
`

`

`UBICOMM 2010 : The Fourth International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies
`
`beginning and end. Moreover this new technology helps
`patients to wear the mobile device at home. This technology
`is the gyroscopes that being used together with the
`accelerometers can enhance the physical rehabilitation.
`In summary, our proposal contributes to the ubiquitous
`health care. Our system improves the physician monitoring,
`guides patients on the rehabilitation process, and can reduce
`the problem of health care systems overcharge.
`
`ACKNOWLEDGMENT
`This work has been financed by PII1I09-0123-27 and
`HITO-09-50 projects from Junta de Comunidades de
`Castilla-La Mancha, and by the TIN2009-14406-C05-03
`project from the Ministerio de Ciencia e Innovación (Spain)
`
`[2]
`
`REFERENCES
`
`[1] Z. Prekopcsk, “Accelerometer based real-time gesture recognition,”
`Proc. International Student Conference on Electrical Engineering,
`Prague, Czech Republic, May 2008, pp. 1-5.
`J. Doscher and C. Kitchin, “Monitoring machine vibration with
`micromachined accelerometers,” Sensors, vol. 14(5), 1997, pp. 33-38.
`[3] M. Joselli and E. Clua, “grmobile: A framework for touch and
`accelerometer gesture recognition for mobile games,” Proc. Brazilian
`Symposium on Games and Digital Entertainment, Rio de Janeiro,
`Brazil, Oct. 2009, pp 141-150, doi:10.1109/SBGAMES.2009.24.
`[4] K. M. Culhane, M. OConnor, D. Lyons, and G. M. Lyons,
`“Accelerometers in rehabilitation medicine for older adults,” Age and
`Ageing, vol. 20, Oct. 2005, pp. 556–560, doi:10.1093/ageing/afi192
`[5] E. Jovanov, A. Milenkovic, C. Otto, and P. C. de Groen, “A wireless
`body area network of intelligent motion sensors for computer assisted
`physical rehabilitation,” Journal on Neuroengineering Rehabilitation,
`vol. 2(6), Mar. 2005, pp 1-10, doi:10.1186/1743-0003-2-6
`[6] E. Mpofu and T. Oakland, Rehabilitation and Health Assessment:
`Applying ICF Guidelines. Springer Publishing Company, 2010, 760
`pp.
`[7] T. ODonovan, J. ODonoghue, C. Sreenan, D. Sammon, P. OReilly,
`and K. A. OConnor, “A context aware wireless body area network,”
`Proc.
`Int. Conf. on Pervasive Computing Technologies
`for
`1-8,
`doi:
`Healthcare, London, UK, Apr.
`2009,
`pp.
`10.4108/ICST.PERVASIVEHEALTH2009.5987
`[8] G. Niezen and G. P. Hancke, “Gesture recognition as ubiquitous input
`for mobile phones,” Proc. Int. workshop on Devices that Alter
`Perception, Sep. 2008, pp. 25-28.
`[9] MovilForum, “Aplicaciones ehealth en la feria movilforum 2009,”
`http://pressoffice.telefonica.com/documentos/Dossier (2010, Feb 20th)
`[10] A. S. Alklind Taylor, P. Backlund, H. Engström, M. Johannesson, H.
`Krasniqi, and M. Lebram. “Acceptance of Entertaiment Systems in
`
`Stroke Rehabilitation,” Porc. IADIS Game and Entertaiment
`Technologies, Algarve, Portugal, Jun. 2009, pp. 75-83.
`[11] T. S. Leong, J. Lai, J. Panza, P. Pong, and J. Hong, “Wii want to
`write: An accelerometer based gesture recognition system,” Porc. Int.
`Conf. on Recent and Emerging Advanced Techologies
`in
`Engineering, Pan Pacific, Malaysia, Nov. 2009.
`[12] R. Niels and L. Vuurpijl, “Using dynamic time warping for intuitive
`handwriting recognition,” Proc. Conference of the international
`graphonomics society (IGS2005), Salerno, Italy, Jun. 2005, pp. 217–
`221.
`[13] L. A. R. Solano, Verificacion del hablante basado en Dynamic Time
`Warping. Universidad del Norte, 1998.
`[14] T. Schölmer, B. Poppinga, N. Henze, and S. Boll, “Gesture
`recognition with a wii controller,” Proc. 2nd international conference
`on Tangible and embedded interaction, Bonn, Germany, Feb. 2008,
`pp 11-14, doi: 10.1145/1347390.1347395
`[15] D. Lopez-de-Ipiña, X. Laiseca, A. Barbier, U. Aguilera, A. Almeida,
`P. Orduña, and J.I. Vazquez. “Infrastructural support for ambient
`assisted living,” Advances in Soft Computing 51, Springer, 2009, pp
`66-75.
`[16] J. Garduño, L. Pedersen, P. Brodal, T. Sund, G. K. Klungsøyr, T.
`Konstali, and H. Gundersen. “OMA-BCAST Quiz Gaming Prototype
`for Mobile”. Proc. IADIS Game and Entertaiment Technologies,
`Algarve, Portugal, Jun. 2009,
`[17] M. García-Herranz, P.A. Haya, A. Esquivel, G. Montoro, and X.
`Alamán. “Easing the smart home: Semi-automatic adaptation in
`perceptive environments,” Journal of Univeral Computer Science vol.
`14(9), 2008, pp. 1529-1544.
`[18] S. Choquette, M. Hamel, and P Boissy. “Accelerometer-based
`wireless body area network to estimate intensity of therapy in post-
`acute rehabilitation,” Journal of NeuroEngineering and Rehabilitation
`vol. 5(20), 2008, doi: 10.1186/1743-0003-5-20
`[19] S. Arteaga, J. Chevalier, A. Coile, A. W. Hill, S. Sali, S.
`Sudhakhrisnan, and S. H. Kurniawan. “Low-cost accelerometry-based
`posture monitoring system for stroke survivors,” Proc. 10th
`international ACM SIGACCESS conference on Computers and
`doi:
`243-244,
`accessibility. Halifax, Canada,
`2008,
`pp.
`10.1145/1414471.1414519.
`[20] M. W. Hinkel. “WiiHab rehabilitative therapy using the Wii.”
`http://wiihabtherapy.blogspot.com . (2008, May 10th)
`[21] ST Microelectronics. “LIS302DL Datasheets and characteristics,”
`www.st.com/stonline/books/pdf/ docs/12726.pdf (2010, May 29th )
`[22] J. Bravo, D. Lopez de Ipiña, C. Fuentes, R. Hervás, R. Peña, M.
`Vergara and G. Casero. “Enabling NFC technology for supporting
`chronic diseases: A proposal for alzheimer caregivers,” LNCS 5355.
`Springer Publishing Company, 2008, pp. 109-125, doi: 10.1007/978-
`3-540-89617-3_8
`[23] R. Hervás, S. W. Nava, G. Chavira, V. Villarreal and J. Bravo.
`“PIVITa: Taxonomy for displaying information in pervasive and
`collaborative environments,” Advances in Soft Computing 51,
`Springer Publishing Company, 2009, pp. 293-301, doi: 10.1007/978-
`3-540-85867-6_34
`
`
`
`
`
`
`
`Copyright (c) IARIA, 2010 ISBN: 978-1-61208-100-7
`
`421
`
`Petitioner Samsung Ex-1048, 0006
`
`