Physiotherapy Exercises at Home:
Using a Mobile Webapp for Sensing and Supporting the
Performance of the Eccentric Calf Muscle Exercise
SUBMITTED IN PARTIAL FULLFILLMENT FOR THE DEGREE OF MASTER OF SCIENCE
Norbert Kuipers
10760903
M
ASTER
I
NFORMATION
S
TUDIES
H
UMAN-
C
ENTERED
M
ULTIMEDIA
F
ACULTY OF
S
CIENCE
U
NIVERSITY OF
A
MSTERDAM
July 15, 2016
Supervisor: 2nd reader:
Dr. J.A.J. Dierx MPH Dr. A. C. Nusselder
2
Physiotherapy Exercises at Home:
Using a Mobile Webapp for Sensing and Supporting the
Performance of the Eccentric Calf Muscle Exercise
Norbert Kuipers
University of Amsterdaminfo@norbertkuipers.nl
Abstract
This thesis describes the design, development and evaluation of a prototype webapp that measures the training parameters of the eccentric calf muscle exercise and supports the user during the performance with audio. This exercise is often prescribed in case of an Achilles tendinopathy injury which is common among athletes especially middle-aged men. Starting point for the design is the use of things that are already in the possession of the user like a smartphone and a running bracelet and from developers perspective the use of web standards in order to create a cross platform app with one codebase.
The sensing focuses on the repetitions in a series and the duration of the repetitions. The duration is an important performance indicator while the exercise is often issued with the instruction to perform the eccentric movement slowly. That is also the reason to focus the design of the audio support on supporting the user to perform the exercise slowly.
The sensing in the prototype is done by using the device orientation event while the smartphone is placed on the foot. The audio support consist of a verbal/voice metronome that counts the repetitions and the seconds of the eccentric movement.
A test (N=16) shows that repetitions and the duration are measured accurate and the majority of the users perform the exercise better regarding duration with the audio support. A questionnaire about the experience and usability results in a positive experience regarding the audio support and less positive (but still positive) experience regarding the usability and operation of the app. Keywords
Mobile computing, sensing, webapp, audio support, eHealth, Achilles tendinopathy, rehabilitation.
1. Introduction
This thesis is the first part of a larger project concerning the use of multimedia tools for increasing therapy compliance in physiotherapy exercises at home. The research in this thesis focuses on the use of a webapp on a smartphone as a sensing and supporting device for a specific physiotherapeutic exercise. The perspective is that in the future this sensing and supporting by the smartphone provides a basis for persuasive applications for improving therapy compliance.
Therapy non-compliance by patients has been recognized as a major problem in prescribed home physiotherapy exercises. Campbell et al [1] discovered that non-compliance with physiotherapy is common and from the patient point of view often a rational decision. Many different causes for the non-compliance are mentioned: lack of time, pain, motivation, forget to do exercise, etc. [2] [3]. Achieving therapy compliance is first of all a behavior change and therefore persuasive computing technologies can be
used as trigger as B.J. Fogg describes in his behavior model for persuasive design (FBM) [4]. For instance notifications can be used in case of forgetting or perhaps gamification and live feedback in case of lack of motivation. Incel et al [5] also mention persuasive computing methods as important challenges for behavior or lifestyle change applications in the context of activity recognition with mobile phones.
Chandra & Oakley [3] formulated six insights and recommendations regarding designing applications for support on prescribed home physiotherapy exercises in order to tackle the problem of non-compliance. Six scenarios are presented showing persuasive concepts for motivation by:
Understanding Enjoyment Results Scheduling Support and Peers
1.1 Smartphone and sensing
In at least one of the described scenarios, understanding where real time feedback is generated, live sensing of muscular activity is done by a custom hardware device [3]. One can imagine that for other scenarios like Enjoyment (e.g. gamification) or support (e.g. remote monitoring by a physiotherapist) live sensed data can be the necessary input. For the user live sensed data has an advantage that there is no need for data entry manually.
But from a user-centered design perspective the use of custom devices for this sensing has some major drawbacks. They need to be purchased or borrowed from a caregiver, are sometimes attached to the body in a difficult way or needs to be calibrated. In physiotherapy devices like the Kinect and Wii are also used for sensing, but they have the same problem of availability. In 2010 Hynes et al [6] already plead for the use of off-the-shelf mobile phones for measuring patient’s activities. The modern mobile phones have many sensors for measuring activity and motion. Incel et al [5] reviewed 36 mobile applications regarding activity recognition and provide a taxonomy with among others used sensors and platforms. Nowadays the smartphones are widely spread and have multimedia possibilities that can be used for persuasive applications.
The mobile phone also has disadvantages like limited data storage and processing power, which can lead to challenges when for instance real time running classifier algorithms are necessary for the application [5]. Depending on the purpose of an application usability issues concerning the operation, placement of phone or visual feedback can occur. For instance where to place the mobile
3 without obstructing the exercises and how to give support or
feedback when the mobile is on one’s ankle for instance? Also the diversity of phones, platforms and sensors can lead to problems.
1.2 Web standards
The last decade the trend has been developing apps with native programming languages like Java for the Android platform and Objective-C for IOS. Since the emergence of HTML5 [7], there are new opportunities for developing mobile applications with web standards (HTML, CSS, Javascript) that work cross-platform within the browser of a mobile. With tools like Cordova and Phonegap a webapp can be transformed into a hybrid app that has a native shell for the different platforms but still the same code base with Javascript [8] [9] [10]. Javascript is typed as a faster-to-learn language. More and more app development companies choose Javascript as the core development language for economic considerations [11]. For the development of our application we hook up with this trend. We use HTML, CSS and Javascript and will especially investigate if the abilities of JavaScript to read the sensors of the mobile phone is sufficient for the intended application.
1.3 Physiotherapy exercises
Physiotherapy exercises are described in so called rehabilitation or exercise protocols. Habets & Cingel [12] reviewed different protocols for the eccentric exercise training. Part of the protocol are
training parameters like:
- Period, often quantified in weeks.
- Frequency per day
- Sets or Series per time
- Repetitions per set.
- Speed, expressed in terms such as ‘low’ and ‘high’ or in
seconds or in counts1.
Depending on the type of exercise many other parameters can be distinguished and also conditions like “Add weight in case of no pain experienced” are described in a protocol [12].
Therapy compliance can be considered as the consequent executing of the exercise protocol issued by the physiotherapist. Measuring the training parameters from the protocol during execution of the exercise by the smartphone delivers data for evaluating the compliance to the protocol and ensures automatic data gathering for the purpose of persuasive design and the possibility of direct support and feedback for the user.
Therefore the first problem (Q1) this thesis addresses is the accuracy of measuring of the training parameters during the performance of exercises.
For this research the eccentric calf muscle exercise with straight knee is used (video example [13], [14]). This is a commonly prescribed exercise in case of a chronic mid-portion Achilles tendinopathy. This exercise has some interesting usability challenges like the placement of the mobile phone. Furthermore this exercise is also interesting while the training in general has been proven effective but there is debate about the optimal protocol which was the motivation for Habets & Cingel to review the different protocols [12]. When in the future a tool is developed that
1 In fact this is more duration than speed when expressed in counts or seconds.
measures the compliance to the protocol this can perhaps be used in research for the optimum protocol.
The training parameters Sets per Time, repetitions and duration from this specific exercise are investigated. These parameters are very common for all kinds of exercises and they deliver a quick indication of compliance to the issued regime. Although the parameters Frequency and Period are also important for compliance measuring they are left out because they are not fundamental different from Sets per time and they would extend the test time substantial.
1.4 Audio support
By measuring the training parameters with the smartphone the multimedia possibilities of the smartphone are not yet used for supporting the performance. B.J. Fogg distinguishes three factors in his Behavioral Model for Persuasive Design (BFM): motivation,
ability and trigger [4]. In this thesis we want to explore the
possibility of enhancing the ability factor in case of physiotherapy exercise by using one of the modalities of the smartphone for direct support: audio. Does the audio support increase the ability? The influence of the audio support on the exercise performance is therefore the second research question (Q2) of this thesis. There is an obvious reason why we choose audio for support. When we want to stick to the smartphone as the only device for live sensing and direct support simultaneously we have to consider that visual support is difficult while the smartphone is put somewhere near the ankle or foot and is out of sight of the subject. The modalities that alternatively can be used are vibration and audio. In this research the use of direct audio support is investigated, in particular the use of audio support in performing the exercise conform the training parameters. When these parameters are also measured at the same time a conclusion can be drawn about influence of the audio support on the performance. When direct support leads to a better performance of an exercise it is a first step towards increased compliance.
1.5 User experience
This research focuses on two basic or initial components of future persuasive applications: sensing performance data by the webapp and direct audio support during performance.
But when the webapp is capable of measuring parameters and the audio support indeed enhances the performance it is still not a guarantee the users will use the webapp. Questions concerning the experience and usability are also important to investigate. How do the users experience the placement of the mobile? Is the operation of the webapp complex? How pleasant is the audio? The user experience and the usability of the webapp is the third research domain (Q3) of this thesis.
1.6 Prototype
Before we can perform tests to investigate the questions we need to develop a prototype with which the test can be conducted. The intention is to develop a user friendly and sustainable prototype that senses the training parameters of the performance of the eccentric calf muscle exercise and that supports the patient in the execution of the exercise with audio.
4 Figure 1: sport bracelet
The prototype is sustainable in the sense of making use of equipment that is already in possession of the target group or easy to achieve. Therefore the test will be conducted with a common smartphone and a commonly used sport bracelet (Figure 1: sport bracelet) put around the ankle or foot. As stated earlier the prototype will be developed with web standards.
The emphasis of this research is on the user and his performance of the exercise conform the protocol training parameters and is not on researching the exact movement for diagnostic or medical analyzing purposes.
2. Research questions
Based on the motivation described in the introduction we distinguished three main research questions:
Q1 What is the accuracy of the sensing of the training parameters
from the eccentric calf muscle exercise performance by a mobile webapp?
To research this question a prototype webapp will be developed that is capable of sensing the specific parameters of the example exercise. Both the exercise and the sensor algorithms for a webapp need to be investigated. One of the sub questions to be answered is whether the placement of the smartphone around the ankle is sufficient for accurate sensing.
Q2 Does audio support of a mobile webapp improve the
performance of the eccentric calf muscle exercise?
To answer this question the prototype needs to be extended with audio support To come to a design the domain of audio support needs to be explored and algorithms concerning the use of audio on smartphones have to be studied.
Q3 What is the user experience of the use of a mobile webapp
with audio support during the performance of a physiotherapy exercise?
The user experience concerns placement of the smartphone, comfort during exercise, the audio support, the interaction with the webapp and the provided information.
3. Related work
Many smartphone applications are designed and developed especially in the context of activity recognition and monitoring for healthcare, personal wellbeing and fitness [5]. People are familiar with apps like Runkeeper [15] easily used by wearing a belt around the arm with the smartphone in it.
Most running and cycling apps are usually measuring exercise parameters automatically mainly by the GPS sensor of the smartphone. Fitness workout apps like Gym Hero [16] and Jefit [17] requires manually input for recording the training parameters.
Audio support is available in several exercise apps in the form of verbal repetition counters or more metronome like with sounds. An example of the latter is the Weight Timer & Trainer app from Tovlife [18]. In this app every repetition is divided into four phases. Every phase has its own sound which plays every second. The user can choose between sounds or voice support. The voice support exists of commands like “hold” and “move” and a count every two seconds to indicate the progress of the current movement. The amount of repetitions is not pronounced and only visual. The voice is very mechanical.
Pernek et al. [19] developed a real time repetition detection algorithm for a broad range of fitness resistance training exercises based on the accelerometer of an Android smartphone. This algorithm also determines the duration of the repetition and gives visual feedback regarding the duration of the performed repetitions. The algorithm is based on similarity matching with prerecorded patterns from the test subjects themselves or training experts. In the context of physiotherapy and rehabilitation research is done regarding the use of smartphones. An example is the study of Galán-Mercant and Cuesta-Vargas who use the inertial sensors of an iPhone for studying and analyzing the Extended Timed Up and
Go test [20]. A study performed by Giggins et al [21] shows that
one inertial sensor can accurately classify exercise performance in patients performing lower limb exercises for rehabilitation purposes, also compared with the Kinect. The application with which Professor Brian Caulfield of UCD, won the 2015 Google Wearables in Healthcare Pilot Challenge [22], has real-time visual feedback on an iPad and some gaming elements. He also proposed a design approach for interactive feedback technology supports in rehabilitation [23]. In this approach he calls for attention for the expectations and needs of both the clinicians and the patients; the clinical challenges must be understood, the possible benefits of technology must be investigated and the usability must be evaluated.
Spina et al [24] developed a training system for an Android smartphone for motion rehabilitation exercises for COPD patients. Their algorithm counts the repetitions and determines the speed and range of motion based on acceleration and orientation data. The data is classified and the system provides the user with real time acoustic feedback about the exercise performance. This system has a teach mode in which the system is personalized. In the applications that are developed in the studies of Caulfield [22] and Spina et al [24] and described above, movements are measured with the aim to provide the user with real-time feedback. Therefore the data must be processed real-time. In our prototype we intend to use audio support that is disconnected from the real performance and therefore there is no need for real-time processing of the sensed data.
In none of the systems or apps the eccentric calf muscle exercise is used. In COPD trainer [24] step ups are measured with the smartphone mounted on the ankle. This exercise yields a relatively large amount of measurement errors.
4. Design and development prototype
This research has features of a problem solving, user-centered, interaction design project. The design process starts with exploring the problem space. The user, the (ex-) patient, the exercise and physiotherapist need to be studied. After this exploration design criteria are determined and in an iterative process a prototype is designed and developed. The prototype is continually tested and adjusted. On regular basis physiotherapists are consulted.5
4.1 Exploration of problem space
A literature review is conducted and two users and two physiotherapist are interviewed in order to understand the exercise, the performance (at home) and identify problems and chances. The exercise is often issued in case of an Achilles tendinopathy. This is a common injury among athletes but there are also physically inactive patients. The patients are of all ages and gender but especially middle-aged men are overrepresented [25]. The exercise exists out of a downward movement (eccentric) and an upward movement (concentric). Sometimes the patient is issued to hold in the lower position. The patient is instructed how and with which parameters the exercise should be performed at home. Most instructions are based on the protocol of Alfredson [26] [12], but the parameters differ very much in times per day, series per time and repetitions per series. 3 times a day, 3 series of 10, 12 or 15 repetitions are very common.
There is a difference in the instruction about the speed or duration of the eccentric movement. The first interviewed physiotherapist (Ph1) often issued one second or count, but mentions that in recent years the literature pleads for slower eccentric movement. Ph2 says: “the eccentric part must be done intentionally very slowly”.2 He orders the patient to count along: “1, 2, 3, 4, 5” and “Most of them tend to move too fast”.
There is also a difference in the way the exercise is performed regarding the use of one or two legs. It depends on how much load on the leg is possible. Often two legs are used in the concentric movement.
The first interviewed user (U1) can’t remember training parameters issued by his physiotherapist, except for ”three times a day”. User 2 (U2) says 3x per day 1 series of 10, but no duration or speed. What also differs is the way people pause in between series. Ph1 urges his patient to do the other leg. U2 has no pause because he always does one series.
U1 mentions notifications, instructional videos, real-time vocal support and information about the training parameters as additional functionalities for the app. He is also interested in feedback about the performance based on the sensed data directly after execution of the exercise and he also would be stimulated by gamification. He likes to be challenged. Finally he also suggests online connection with fellow injured persons. Ph2 uses the smartphone of the patient to make a video of a good performance of the exercise for reference at home.
U2 indicates he likes the idea of an app that supports and assists. Besides notifications in case of forgetting to do the exercises, he likes some information and confirmation in the right performance of the exercise. He mentions explicit evaluation of the depth of the movement of the heel.
U2 and both the physiotherapists suggest an online connection with the physiotherapist and like the idea of audio support in counting. Ph2 expresses his doubt about the placement of the smartphone on the foot or ankle. His doubt concerns the motivation for patients to do the effort and he states that a valid measuring of angles is not possible. He doesn’t believe in qualitative judgement of the performance by a smartphone regarding the exact movement. For sensing of less complex issues like amount of performed series the smartphone is suitable in his opinion.
2 The interviews were held in Dutch. Citation are translated by the author.
4.2 Design criteria
A set of criteria for the prototype design is derived from the motivation, the literature review and interviews:
- Functionality focus on sensing and supporting the exercise variant with the slow performance of the eccentric movement.
- The prototype must measure the amount of repetitions in a series and the duration of an individual repetition (in fact the duration of the eccentric phase). Due to the unpredictable way of pausing between series for now only the individual series needs to be stored with date and time so afterwards the number of series on a day can be calculated.
- The interaction design must take into account the extraordinary position of the smartphone. - The prototype is developed with web standards
- The aim is to develop simple algorithms that limits the load on the processor and demands no training of the system or calibration by the user.
4.3 Algorithm
A preliminary version of the prototype is developed that senses the data produced by the device orientation and device motion events [27] and stores the data into a CSV file. Both events deliver data in three dimensions. Device orientation shows angles in degrees, device motion gives acceleration in m/s2. The data is analyzed. Figure 2 shows an example of the sensed data. The devicemotion data is much noisier than the deviceorientation data.
Figure 2: Sensed data from exercise performed with phone placed on lower leg near ankle.
After several tests with applied filters and transformation we didn’t succeed in producing an acceptable peak detection algorithm with the devicemotion data and we decided to use the deviceorientation data.
To develop the algorithm a desktop version of the prototype is developed in which the raw data CVS files can be loaded. In this way the algorithm can be fine-tuned based on earlier performed exercises. Several performances were video captured in order to compare the real performance with the results produced by the algorithm.
The algorithm developed is based on the observation that the mobile placed on the foot or lower leg makes an angle relative to an axis of the starting position during the movement (Figure 3). High peaks and low peaks indicates turning points in the movements.
6 Figure 3: angle change detected by deviceorientation event
during movement.
The algorithm to process the sensed data is visualized in Figure 4. During the exercise the data is captured with a frequency of 100 Hz. This is rather high, but because the processing of the data is done after the stop this high frequency does not lead to performance problems of the processor. The graphs in Figure 2 show distortion towards the end. This distortion turned out not to be caused by performance problems but the sleep mode of the mobile. The mobile is placed in de bracelet and sometimes the light sensors are covered causing the mobile turning into sleep mode after a set period. Now the algorithm contains a “nosleep” command.
Figure 4: the algorithm for processing the data
Every data point exists of the angle and a timestamp. After the stop the raw data is stored and the processing starts. In a few tests the graph shows a remarkable offset of 180 degrees (Figure 5). It is not clear why the offset occurs, but it needs to be corrected before the data is further processed. A correction filter is built to detect the offset and correct it.
Figure 5: graph with offset
To smoothen the data a moving average filter is applied. The range of the moving average is determined on 80 based on Spina et al. [24]. The next step is determination of the direction and peak detection based on change of direction. Despite the moving average filter there are still direction changes that turn out to be not real peaks, especially in the neighborhood of a turning point in the movement. The peak detection algorithm only identifies a peak when the n previous data points are in the same opposite direction of the peak data point. n depends on the frequency. Finally the peak detection checks on double peaks or missed peaks. When two succeeding peaks in the same direction are detected within a threshold the algorithm classifies them as double peaks and deletes
3 For the test this measure time is taken out of the settings.
the peak with the lowest deviceorientation degree value. When the two peaks are not within the threshold the algorithm determines a missed peak and add a peak in the other direction in between the two peaks of the same direction. The threshold depends on the set intended duration of the movement.
The peaks are turned into repetitions of the eccentric movement. Based on the timestamp of the begin peak and the end peak the duration of a repetition is determined. Sometimes a user does not start from the horizontal start position but in the first top position. The identified repetitions are in that case the concentric movements. The average duration of all the repetitions and the opposite movements are calculated and compared. Based on the premise that the eccentric movement should be performed slower than the concentric movement the repetitions are mirrored if necessary.
Some users make a small up and down movement half way the movement. Based on the set duration of the eccentric movement small repetitions are merged with the previous repetition.
From the intended number of repetitions (stored in the settings) minus one the algorithm looks for small or large (compared with the average of the previous performed repetitions) repetitions to determine the end of a series. Finally the average and standard deviation is calculated and display. The repetitions are stored in a CSV file. In a new version of the app the data will be stored in the local storage of the phone and used for determining the series per day.
4.4 Design
Interaction
The user starts the app by pressing the start button (Figure 6). In the settings the user can set a waiting time to start. The waiting time is meant to give the user the opportunity to put the mobile in the bracelet after pressing start. The start button is made very large so the user can choose to put the mobile in the bracelet and then press start.
After pressing start the app starts a voice countdown to create a known starting point of measuring. Depending on the settings the audio support is started after the countdown. The start button changes into a stop button.
When the user is ready he can press the stop button, but he can also wait until the measure time3 is over.
7
Information
After the app is stopped the average duration as an indication of the performed tempo and the standard deviation as an indication of the performed rhythm is presented. Also a table is displayed with all the performed repetitions with the individual duration and differences to the norm (the issued duration from the settings). In an earlier version also the performed angle was displayed in the table. The measured angle depends on the placement of the mobile and is therefore not a valid indication of the real performed angle by the user. We decided not to present this information and only use the angles for peak detection.
Audio support
During tests we observed that users encounter problems in counting the repetitions and at the same time counting for the duration of the eccentric movement. The audio support is therefore designed as a verbal/voice metronome that not only counts seconds but also repetitions. The audio counts “21, 22, 23, 24” and then says “one” for the first performed repetitions and so on till the last repetitions is done according the settings. The assumption is that verbs have additional information above sounds and therefore relieves the user more in his cognitive tasks. The audio is not connected to the real performed repetitions. The idea is to provide the user with a rhythm he can hook in.
For the tests an audio fragment is composed for 10 repetitions of a four second eccentric movement and one second upward movement. The intention is to make audio fragments for all common combinations of durations and number of repetitions. To have a pleasant experience a real recorded voice is used instead of text-to-speech technology.
Placement
The prototype was initially developed for placement on the lower leg. Consulting a physiotherapist during prototype iterations learns that when the patient performs the exercise correctly his lower leg doesn’t make an angle. Although all the test persons so far had made angles with the lower leg we deem it fundamental wrong to base the measurement on a wrongly performed exercise.
The foot makes a clear angle when the exercise is performed right. We decide to use this placement and ask the users in the test about the experience. For the algorithm it means we use the Beta of the deviceorientation data instead of the Gamma.
5. Methods
With the prototype a test can be performed to answer the research questions. The research strategy is a mixed approach of qualitative and quantitative methods.
Test
The quantitative part exists of a test with the prototype and a test group of 16 participants performing two series of 10 repetitions. One series is with audio support (T+AS) and one is without audio support (T-AS). To reduce the influence of an eventually training effect (the second performed series is better) the half of the test group starts with T+AS and the other half with T-AS.
The tests are video captured. The videos are viewed and all repetitions are identified and quantified. This data forms the ground truth and are compared with data produced by the prototype. The accuracy of sensing (Q1) is researched with this test. The test is limited to the parameters repetitions and the duration of the repetitions. The data provided by the prototype are compared with the data retrieved by the video observation. The accuracy will be expressed in statistical values.
The influence of the audio support on the performance of the exercise (Q2) is determined by comparing the results of T+AS and T-AS of the participants. The performance parameter that is tested is the duration of the repetition. We hypothesize that the series performed with audio support are performed more evenly regarding the duration (rhythmic) and closer to the imposed duration (right tempo).
Questionnaire
The experience of the user (Q3) is evaluated with a questionnaire (appendix A) with mainly qualitative questions. The questions are a mix of Likert scale and open questions and are about the audio support, the information provided by the app and the operational process of the app including the placement of the mobile. The formulation of the questions is inspired by the System Usability Scale (SUS) [28].
Participants
The including criterion is experience with the eccentric calf muscle exercise. The test persons must be familiar with the exercise and perform the exercise regularly (or have performed) in connection with recovery, for example, imposed by
physiotherapy. This group has experience and can therefore relate to the test and the questions to his/her own situation. In addition, it may occur within this group that the participant is somewhat obstructed in the performance of the exercise is by the injury. Also in this situation should be tested.
The app is ultimately intended for home use. Therefore, preference is given to the test at home. But for logistical reasons test at work or in a physiotherapy practice is also suitable. Especially since the exercise is in practice also carried out on the two latter locations. Participants are recruited from staff and students of the Avans University Applied Sciences and the Athletic club in Breda in the Netherlands. Table 1 shows the test group.
Table 1: Participants of test
nr Gender Age Location
1 f 23 School/work 2 m 44 School/work 3 m 57 School/work 4 m 39 home 5 m 46 School/work 6 f 50 School/work 7 m 58 Athletic club 8 m 48 Athletic club 9 m 48 Athletic club 10 m 58 Athletic club 11 m 50 home 12 f 36 School/work 13 m 47 home 14 m 50 School/work 15 f 41 Athletic club 16 f 45 home Material
The test is performed with a Samsung Galaxy S4 mini and the Chrome browser.
Interview with physiotherapists
Finally the prototype is demonstrated to the same physiotherapist who were interviewed during the exploration of the problem space. With an open interview open-ended feedback will be retrieved.
8
6. Results
6.1 Accuracy of the sensing of the training
parameters
Table 2 shows the number of repetitions in a performed series compared with the number of video observed repetitions.
Table 2: the number of repetitions in series measured by app and observed on video.
Participant #reps T-AS app #reps T-AS video #reps T+AS app #reps T+AS video 1 12 12 10 10 2 10 10 11 10 3 10 10 10 10 4 10 10 10 10 5 10 10 11 11 6 13 12 11 11 7 10 10 10 10 8 6 8 6 6 9 8 8 12 10 10 11 11 8 7 11 10 10 11 10 12 10 10 11 11 13 12 12 10 10 14 11 11 10 10 15 10 10 11 11 16 10 10 10 10
From the 32 executed series 26 are counted correct. Analysis of the video and the data learns that all miscounts (2, 6, 9, 10 and 11), except for T-AS from participant 8, concern the last movement of the subject that is not intended to be a repetition but is to get into the stop position. Those movements are in fact large enough to be recognized as a repetition by the app. In case of participant 9 T+AS the last movements are even identified as two small repetitions while the mobile is put out of the sport bracelet before stopping the app. When analyzing the video sometimes it is hard to see whether or not the last movement is an intended repetition or not.
Participant 8 took two to three seconds for the concentric movement out of caution because of a serious injury. For this reason he made only 8 repetitions in 60 seconds and the measurement was then cut off by the timer of the app. The app missed the first and the last repetition. The 6 middle repetitions were identified correctly.
In total for T-AS 164 repetitions and for T+AS 157 repetitions were observed by video. From that 321 repetitions only 2 in total were missed by the app and there were 6 movements registered by app but not by video.
We consider the measuring of the amount of repetitions remarkable accurate when it concerns the identification of the individual repetitions. The handling of the last repetition by the algorithm needs another rethinking and redesign.
Duration
For determining the accuracy of the measurement of the duration of the eccentric movement all the repetitions determined by the app are taken into account that are also identified as such by the video observation. Only one additional repetition is filtered out because it has an extreme value. This is the last repetition of T-AS of test subject 8 (P8) determined by the app. Due to earlier mentioned
reason this test was cut off by the timer causing an odd duration value and is excluded from further analysis. Due to a file storage error the individual repetitions (N=10) of T-AS of participant 15 were not stored and cannot be used for a comparison of the duration with the video observed repetitions. Therefore the cleaned dataset consists of the repetitions that are identified by the video minus the last repetition of T-AS of P8 and minus T-AS of participant 15. For all repetitions of this cleaned dataset (N=308) the difference in duration is calculated with the corresponding video observed repetition duration. The mean difference is -0.13 seconds and the mean distance (absolute values of the differences) is 0.19s. The average of all video observed repetitions is 3.51s. This means that accuracy of measuring of the duration by the app compared with the video ground truth is 95% if we look at the mean distance. What is striking about the data distribution is that it is not symmetric and has a negative skewness. Figure 7 shows that most of the durations measured by the app are shorter than the video observed duration.
6.2 Audio support and performance
For analyzing the influence of the audio support the average and standard deviation durations of the repetitions of the performed series in T-AS and T+AS and measured by the app are compared. The cleaned dataset defined in 6.1 is used again, but because the average and standard deviation of the series are used the T-AS results of test person 15 can be included for this analyses while the average and standard deviation were recorded on paper during the test for this test.
Before we can determine if the performance improves by the audio support we have to rule out a significant training effect. A training effect can occur for the second performed test because one has already done a series just before. Especially when the first test was T+AS one can have the audial counting “in mind”. We performed an independent-Samples T test for variables T-AS average, T-AS
standard deviation, T+AS average and T+AS standard deviation.
The grouping variable is the first variable in which the first performed test by the subject is identified (T-AS or T+AS).
Figure 7: N=308 repetitions
Positive score means the duration measured by the app is longer than the corresponding video observed repetition.
The repetitions in the zero bin are equal measured by the app and video observed.
Negative scores means the duration measured by the app is shorter than the corresponding video observed repetition.
0 20 40 60 80 -0 ,6 -0 ,5 -0 ,4 -0 ,3 -0 ,2 -0 ,1 0 0,1 0,2 0,3 0,4 0,5 0,6 # re pet it io ns diff in s
9 Table 3: Independent-Samples T tests with the first performed
test as grouping variable. First N Mean Mean
diff
Sig. (2-tailed) T-AS average T-AS 8 3.04 -.27 .513
T+AS 8 3.31
T+AS average T-AS 8 3.44 -.48 .135 T+AS 8 3.92
T-AS Stdev T-AS 8 0.33 .039 .532 T+AS 8 0.29
T+AS Stdev T-AS 8 0.30 -.48 .114 T+AS 8 0.43
Table 3 shows the results of the independent-samples t-tests. The t-tests shows that for all four variables the differences are not significant.
The test group is rather small for a conclusion about the training effect because the test group is divided into two groups for this sub test. For a detailed research towards training effect the test group should be larger and another test setup is needed. For now we deem important to rule out a significant training effect in this test in order to use the results of the whole group for investigating the influence of the audio support.
While there is no significant training effect we compare all the T-AS and T+AS pairs on measured average and measured standard deviation in a paired t-test to determine if there is a significant difference between tests without audio support (T-AS) and with audio support .(T+AS). Table 4 shows the results.
Table 4: Results paired t-test to determine influence of audio support. N=16 Mean T-AS Mean T+AS diff Sig (2-tailed) Average duration series 3.17 s 3.68 s 0.51 0.014 Standard deviation duration series 0.31 s 0.37s 0.06 0.359
There is a significant (p=0.014, α=0.05) difference between average duration of the eccentric movement in a performed series of T-AS and T+AS. Audio support leads to a duration that is 0.51 closer to the 4 seconds norm. The standard deviation is worse for audio supported series but this is not significant (p=0.359, α=0.05). As we determined the average as an indication of the tempo of the performance and the standard deviation as an indication of the rhythm we can conclude based on this test that audio support improves the performance regarding the tempo. From the 16 participants 75% performed T+AS slower as T-AS regarding the average duration. 63% performed the T+AS closer to the 4s seconds norm on average duration. This difference is caused by the two participants (P8 and P10) who performed their T+AS in respectively 5.0s and 5.3s and their T-AS respectively in 4.6s and 3.5s. They performed their T+AS although slower than T-AS but also further away from the 4s norm. Regarding to the standard deviation only 31% of the participants performed in a more steady rhythm (=lower standard deviation) with audio support.
As we look at the performed amount of repetitions compared to the issued amount (10) we cannot distinguish a substantial difference: for T-AS 7 participants performed a different amount of repetitions to 6 participants for T+AS.
6.3 User experience and usability
Audio support
In the questionnaire the participants were asked how they experience the audio support. Figure 8 shows that the majority of the participants (13/16) evaluated the audio support positively.
Figure 8: experience audio support
Half of the participants indicated they will probably use the app if it was for free. Another three participants react positively on this question but only for use in the beginning of the exercise period. Five participants indicated not to use the app. From that five participants four performed better on the T-AS than on T+AS regarding the average duration closer to the norm. This could be an indication that these participants have sufficient assets of their own to stick to the intended tempo. This group of five participants contains also the three participants who react neutral or negative on the experience of the audio support. The explanation of these three participants on the negative assessment of the app and audio support are all going in the same direction. P9 remarks “find it pleasant to determine my own rhythm”, P10 “I do it more by feel” and P14 “I don’t need it”. P2, who liked the audio support but indicated not to use the app stated: “No, unless combined with multiple exercises”.
Participants who liked the audio and indicate they want to use the app respond on the audio: “Nice that I do not have to count (total amount, number of counts)” (P3) and “convenient and brings discipline” (P7). Participants who liked the audio and indicate they want to use the app only in the beginning want to learn the tempo or rhythm: “Once the rhythm is sufficient” (P6) and “to teach myself the time interval a "podcast" is sufficient” (P11).
P2 and P14 want more time for the upward movement and P15 suggest in this respect an option in the settings for the duration of the upward movement. Also the volume is an issue for some participants. P8 performed the test in a noisy surrounding. P13 suggest a volume control on the display and P10 the use of headphones. P11 suggests the use of sounds for the counts and voice for the amount of repetitions. P13 suggests another voice. The questions about improvement of the audio delivers a set of useful suggestions. P2 and P14 complain about the short time for the upward movement. P3, P4 and P6 ask for more verbal help at the beginning or the end. P8 asked for more volume.
Provided information
In the questionnaire three questions (Q9, Q10 and Q11) were about the provided information. These questions were not filled in by P10 for unknown reasons. So n=15 for this set of questions.
0 2 1 5 8 0 5 10 F re qu ency
Experience audio support
Not pleasant
10 Figure 9: Opinion about information provided by app.
1=totally disagree, 4=totally agree.
Figure 9 shows the responses to the statements on the information provided by the app. q9 is the average duration of the repetitions, q10 the standard deviation and q11 the table with the duration of all individual repetitions. In general the participants appreciated the provided information but the standard deviation, which reflects the steadiness of the rhythm, is appreciated less.
Usability
In the questionnaire the participants were asked what their opinion was about the operating process off the app from beginning to end including putting the mobile on the foot. Figure 10 shows the result of a Likert scale question about how difficult/easy the whole process the participants evaluate.
With a mean of 3.7 on a 1 to 5 Likert scale the operating process is judged slightly positive.
Figure 10: opinion about control process app
Only one participant partly disagreed on the statement “the app is user friendly” while five choose partly agree and nine choose total agree. The improvement suggestions made in this context are mostly about starting and stopping. P3 wrote “make stop button other color (clear to see that the app is launched, without glasses I cannot read it)”. P2 suggest “remote control” and P11 “voice controlled start/stop”. P6 make several suggestions for start and stop control: “…..or automatically launch when you go up or start moving. - Or if you just let the app count 21-22 and then measure as soon as the motion is going on”. P16 wrote “easier on and off. Feedback if you click on stop”.
Through observation we also saw difficulties concerning starting and stopping. The start button is not always activated by the touch through the plastic cover. When the start button is touched before the smartphone is put into the cover the right corner back button is sometimes accidentally touched causing the webapp disappearing. And finally when the stop button is touched the feedback is not always sufficient causing people touching the button again while it has turned into a start button again. The app starts again and the series information disappears. Training and intensive use will perhaps overcome some of the start/stop problems, but this test
results and the identified problems surrounding the measurement of the last performed repetition require a redesign of the start and stop procedure.
Eleven participants totally agreed on the statement “the belt around the foot is not obstructive” and another three partly agreed. Only two participants disagreed. P5 who partly agreed remarks: “easier attachment to the shoe and simplify operation (=awkward to let stick out partly from case)”.
In the other remarks section of the question P2 makes a note about the general design: “I'm looking forward to the more elegant implementation”. P9 discussed the effect the whole experience can have to people: “Placebo effect of attention will be large. Especially in groups where social environment is important”.
6.4 Feedback physiotherapists
The app was demonstrated to the same two physiotherapists who were interviewed in the initial research. Afterwards an open interview was conducted to gain feedback on the app.
Both physiotherapists see added value for the audio support while it helps people to perform the exercise into the right slow tempo. Ph2: “.. counts the number of repetitions and 21, 22, 23, 24, I think it sounds logical. And that you can configure that, so that, I think is practically useful”.
Also the provided information is important: Ph1: “Especially if you are informed about your results afterwards”. The information should be displayed in a different way, for instance with smileys. Both physiotherapist deem (average) duration of the eccentric movement important information. Ph1 suggests to calculate the total performed time under tension over a period while that is the most important goal of the exercise. The physiotherapist doubts about usefulness of displaying the standard deviation.
Ph1 is probably interested in the measured angle if it can be used to give feedback. Ph2 doubts about that and wants the app to be more like a training support tool.
Regarding to the placement of the smartphone on the foot Ph1 thinks it is not a big issue for a patient. He warns for patients to bend while executing the exercise. Ph2 evaluates the placement on the foot as an important disadvantaged from usability perspective. If it is technically possible he would like to see the mobile to be placed in the trouser pocket.
6.5 Evaluation of the algorithm
The stored raw data files from the performed tests were imported and analyzed by the desktop version of the app. All the programmed filters and correction were applied at least once. The data of T+AS of P11 showed the graph with offset and the correction was automatically applied.
7. Discussion and future work
We are aware of the limitations of this research. The number of participants is small for quantitative research, but additional qualitative research with the questionnaire and interviews with the physiotherapists makes the results valuable for discussion and conclusions. The formulated three main research questions will be discussed separately.
Q1 What is the accuracy of the sensing of the training parameters
from the eccentric calf muscle exercise performance by a mobile webapp?
We conclude that the identification and count of the repetitions works quite well for this test configuration. The only substantial
1 1 2 11 2 0 6 7 2 1 2 10 0 5 10 15 1 2 3 4 Fr e q u e n cy
Opinion about usefulness information
q9 q10 q11 0 3 3 6 4 0 5 10 Fr e q u e n cy
Opinion about operating process
Difficult/ Cumbersome
Easy/ simple
11 mismatches are concerning the last movement of a series that is
identified by the app as a new repetition. We suggest to calculate the average (and eventually standard deviation) over the intended amount of repetitions to avoid the influence of a measured extra repetition that is in fact the stop movement.
Duration is measured very accurate except for the fact that the app seems to measure the majority of the repetitions shorter than the video observed duration of repetitions. Most likely this is caused by the moving average filter or the peak detection algorithm. The suggestion will be to make a correction factor in the algorithm.
The test is performed with one type of mobile phone and with one set of parameters. The prototype should be tested in a more robust way on different kind of platforms, mobile types and with different parameters.
Q2 Does audio support of a mobile webapp improve the
performance of the eccentric calf muscle exercise?
We expected that the audio support improves the performance. Concerning the average duration, as an indication of the tempo, we approve the hypothesis, but for the standard deviation, as an indication for a steady rhythm we reject the hypothesis.
Although the sample size of the test is rather small combined with the positive feedback regarding the audio support in the
questionnaire we conclude the audio support can improve the performance for a substantial part of the target group. But a smaller group doesn’t appreciate the audio support. A future research question could be if this disapproval of this multimedia support (negative) correlates with therapy compliance.
The audio support can be improved in several ways. The duration of all the separate stages of the movement must be configurable in the settings, and perhaps an option between voice/verbal and sounds. From technical point of view a new research is proposed to investigate if the audio support can be composed from single audio fragments during the execution of the exercise without delay.
Q3 What is the user experience of the use of a mobile webapp
with audio support during the performance of a physiotherapy exercise?
The audio support is reviewed as positive by both the users and the therapists. The information provided by the app is also rated as positive but the standard deviation as an indication as for a steady rhythm is less appreciated and can be omitted. The overall operation of the webapp is also rated positive but some critical remarks are made.
Related to the earlier mentioned problem of the last movement the whole “end detection” and stop procedure needs to be redesigned. In the current design the measuring stops either when the user hits the stop button or the preset measure time is passed. From the test we conclude that the stop button is not user friendly. We recommend to implement an automatic stop based on preset parameters (including the pause in between series). The user only needs to start the app perform a series within a timeframe based on the settings. The user gets an audio signal when the measuring holds and pauses or stops after the preset amount of series.
Although in this research the placement of the mobile on the foot is not been evaluated as inconvenient one can imagine that it will be
when the app is used in the long term. The placement on the foot is only necessary for automatic measuring. The user must experience added value to persist in doing the effort of placing the mobile on the foot. Because of the concerns about the usability of the placement of the mobile a research towards the performance expectancy and the effort expectancy of the target group conform the Unified Theory of Acceptance and Use of Technology (UTAUT) model is advised [29].
Besides addition of functionality research is needed on how scalable the algorithms are towards other rehabilitation exercises in order to increase usefulness for the user. Other exercises perhaps means other placement on the body and a renewed look at the used sensors. Future developments regarding mobile sensor technology must be monitored.
Web standards
The use of Javascript has certainly its advantages. It is easy to use and has its promise of platform independency. An important disadvantage is that in the scientific world it is not used much. It is difficult to find libraries for filters. The prototype is a webapp and needs to be loaded manually in a browser with all kinds of disadvantages like the need for an Internet connection and accidentally use of the back button. Most of these problems can be solved when the webapp is transformed to a hybrid app. The app can then be used as a standalone native app with still the same Javascript code base. Important conclusion of this research is that web standards can certainly be used for the development of these kind of applications were the sensors of a smartphone are used to measure body movements.
Graphic and audio design
In this project only a little attention is payed to the graphic and audio design. In further development this must be an important issue while it increases the acceptance and usability.
Now we have found that automatic measurement with a webapp for this exercise is possible and audio support can help we need to design new functionalities regarding therapy compliance that make use of the sensed data. Ideas for new functionalities can be derived of the scenarios from Chandra et al. [3]. One can think of notifications, feedback on performance over time and perhaps gamification. A deeper research of the target group is necessary to determine which functionalities improve the motivation and ability and which triggers are needed conform Fogg’s behavior model [4].
8. Acknowledgements
The author would like to thank the supervisor John Dierx for the support and the valuable feedback during the master thesis. I also would like to thank the physiotherapists of Avans University of Applied Sciences, the research group Living in Motion and the Academische Werkplaats Fysiotherapie and all the participants. And last but not least my family who supported me tremendously.
12
9. References
[1] R. Campbell, M. Evans, B. Tucker, B. Quilty and J. Donovan, "Why don’t patients do their exercises? Understanding non-compliance with physiotherapy in patients with osteoarthritis of the knee," Vols. ;55:2 132-138, 2001.
[2] E. M. Sluijs, G. J. Kok and J. van der Zee, "Correlates of exercise compliance in physical therapy," Physical therapy, pp. 73(11), 771-782, 1993.
[3] H. Chandra and I. S. H. Oakley, "Designing to support prescribed home exercises: understanding the needs of physiotherapy patients," in 7th Nordic Conference on Human-Computer Interaction: Making Sense Through Design, 2012.
[4] B. J. Fogg, "A behavior model for persuasive design," in 4th international Conference on Persuasive Technology, 2009.
[5] O. D. Incel, M. Kose and C. Ersoy, "A Review and Taxonomy of Activity Recognition on Mobile Phones," BioNanoScience, vol. 3, no. 2, pp. 145-171, 2013.
[6] H. Hynes, W. M, E. McCarrick and L. Kilmartin, "Accurate monitoring of human physical activity levels for medical diagnosis and monitoring using off-the-shelf cellular handsets," Personal and Ubiquitous Computing, p. 15(7), 2011.
[7] "HTML5," W3C, 28 october 2014. [Online]. Available: https://www.w3.org/TR/html5/. [Accessed 22 june 2016].
[8] "Native, HTML5, or Hybrid: Understanding Your Mobile Application Development Options," [Online]. Available: https://developer.salesforce.com/page/Native,_HTML5,_or_Hybrid:_Understanding_Your_Mobile_Application_Development_O ptions. [Accessed 22 june 2016].
[9] J. Bristowe, "What is a Hybrid Mobile App?," Telerik Developer Network, 25 march 2015. [Online]. Available: http://developer.telerik.com/featured/what-is-a-hybrid-mobile-app/. [Accessed 22 june 2016].
[10] S. Dhillon and Q. H. Mahmoud, "An evaluation framework for cross-platform mobile application development tools: AN EVALUATION FRAMEWORK FOR CROSS-PLATFORM MOBILE APP DEVELOPMENT," Software: Practice and
Experience, vol. 45, no. 10, pp. 1331-1357, 2015.
[11] B. Ray, S. Schuermans and G. Anadiotis, "State of Developer Nation Q1 2016," 17 march 2016. [Online]. Available: https://www.developereconomics.com/reports/developer-economics-state-of-developer-nation-q1-2016#!.
[12] B. Habets and R. E. H. Cingel, "Eccentric exercise training in chronic mid‐portion Achilles tendinopathy: A systematic review on different protocols," Scandinavian journal of medicine & science in sports, vol. 25, no. 1, pp. 3-15, 2015.
[13] Research, Journal of Foot and Ankle, "Eccentric calf muscle exercises for Achilles tendinopathy Part 1, Journal of Foot and Ankle Research," 08 february 2010. [Online]. Available: https://youtu.be/PwMebcSUdgw.
[14] "Eccentric calf muscle exercises for Achilles tendinopathy Part 2, Journal of Foot and Ankle Research," 08 february 2010. [Online]. Available: https://youtu.be/M6EKuuZ7C2E.
[15] "Runkeeper," 31 january 2016. [Online]. Available: https://runkeeper.com/. [Accessed 31 1 2016].
[16] "Gym hero," [Online].
[17] "Jefit," [Online]. Available: https://www.jefit.com/.
[18] "Tovlife," [Online]. Available: http://www.tovlife.com/home.
[19] I. Pernek, K. A. Hummel and P. Kokol, "Exercise repetition detection for resistance training based on smartphones," Personal and
Ubiquitous Computing, vol. 17, no. 4, pp. 771-782, 2013.
[20] A. Galán-Mercant and A. I. Cuesta-Vargas, "Differences in trunk accelerometry between frail and non-frail elderly persons in functional tasks," BMC research notes, vol. 7, no. 1, p. 100, 2014.
[21] O. M. Giggins, K. T. Sweeney and B. Caulfield, "Rehabilitation exercise assessment using inertial sensors: a cross-sectional analytical study," Journal of neuroengineering and rehabilitation, vol. 11, no. 1, p. 158, 2014.
[22] "UCD connected health researcher wins 2015 Google wearables in healthcare pilot challenge," 28 april15 2015. [Online]. Available: http://www.ucd.ie/news/2015/04APR15/280415-UCD-connected-health-researcher-wins-2015-Google-wearables-in-healthcare-pilot-challenge.html.
[23] O. M. Giggins and B. Caulfield, "Proposed Design Approach for Interactive Feedback Technology Supports in Rehabilitation," in
REHAB '15 Proceedings of the 3rd 2015 Workshop on ICTs for improving Patients Rehabilitation Research Techniques, Lisbon,
13 [24] G. Spina, G. Huang, A. Vaes, M. Spruit and O. Amft, "COPDTrainer: a smartphone-based motion rehabilitation training system
with real-time acoustic feedback," in UbiComp '13 , 2013.
[25] J. J. Kingma, R. d. Knikker, H. M. Wittink and T. Takken, "Eccentric overload training in patients with chronic Achilles tendinopathy: a systematic review," British journal of sports medicine, vol. 41, no. 6, 2007.
[26] H. Alfredson and J. Cook, "A treatment algorithm for managing Achilles tendinopathy: new treatment options," British journal of
sports medicine, vol. 41, no. 4, pp. 211-216, 2007.
[27] "DeviceOrientation Event Specification," W3C, 1 12 2011. [Online]. Available: https://www.w3.org/TR/orientation-event/.
[28] J. Sauro, "Measuring Usability With The System Usability Scale (SUS)," Measuringu, 2 febuary 2011. [Online].
[29] V. Venkatesh, M. G. Morris, G. B. Davis and F. D. Davis, "User Acceptance of Information Technology: Toward a Unified View,"
14
Appendix A Questionnaire (in Dutch)
De gegevens uit deze test en vragenlijst worden anoniem verwerkt. In de resultaat beschrijving worden geen namen genoemd. Door onderzoeker in te vullen:
Nr.: Datum test: Tijd test: Locatie:
Volgorde tests T-AS first/T+AS first
Resultaten tests:
T-AS T+AS (met audio)
Gem. duur beweging
Standaardafwijking
Aantal herhalingen
Door participant invullen:
Naam:
Geslacht:
m/v
Leeftijd:
Aangedane (blessure) been
Links/rechts
In deze vragenlijst wordt met ‘beweging’ de excentrische beweging naar beneden bedoeld.
15 De eerste vragen gaan over hoe u de oefening normaal gesproken uitvoert.
1. Hoe lang doet u de kuitspier oefening al? (in weken).
………..2. Welke instructie heeft u meegekregen (van bijvoorbeeld de fysiotherapeut) ten aanzien van aantal keer
per dag, aantal series en herhalingen m.b.t. het doen van de oefening?
(Voorbeeld: 2x per dag 3 series van 15 herhalingen)
….. x per dag …… series van …... herhalingen per serie
Anders: ……… en ten aanzien van de duur van de beweging naar beneden?
(Voorbeeld: 4 tellen)
…… tellen/seconden
Anders: ………
3. In hoeverre lukt het om de oefening volgens deze instructie “trouw” (regelmatig en consequent) uit te
voeren?
O goed O Redelijk goed O matig O slecht
4. Hoe pauzeert u tussen de series?
O Ik wacht op de plaatsO Ik doe het andere been O ik pauzeer niet O Anders:
………. ………. De volgende vragen gaan over de tests die u zojuist met de smartphone om de voet hebt uitgevoerd.
5. In welke mate werd u tijdens het uitvoeren van de oefening belemmerd (door bijvoorbeeld pijn, stijfheid
of vermoeidheid)?
O Helemaal niet belemmerd O Een beetje belemmerd O Redelijk belemmerd O Behoorlijk belemmerd
16
6. In één van de twee testjes werd u via de smartphone met een tellende stem ondersteund (audio
support) bij de uitvoering. Vindt u de audio support prettig?
O
O
O
O
O
Niet prettig 1 2 3 4 5 prettig
7. Stel dat de app met audio support kosteloos beschikbaar is voor uw eigen smartphone, zou u deze dan
gebruiken?
O Ja, waarschijnlijk wel O Ja, maar alleen in het begin O Nee, waarschijnlijk niet O Weet niet Toelichting: ……… ……… ……… ……..
8. Welke verbeterpunten ten aanzien van de audio support kunt u noemen?
……….. ……….. ……….. Na het stoppen van de meting toont de app informatie. Over die informatie gaan de volgende 3 stellingen:
9. “De gemiddelde duur van de beweging vind ik nuttige informatie”
O Helemaal oneensO Een beetje oneens O Een beetje eens O Helemaal eens
10. “De standaardafwijking van de duur van de beweging vind ik nuttige informatie”
O Helemaal oneensO Een beetje oneens O Een beetje eens O Helemaal eens
17
11. “De tabel met de duur van iedere afzonderlijk beweging vind ik nuttige informatie”
O Helemaal oneens O Een beetje oneens O Een beetje eens O Helemaal eens
De volgende vragen en stellingen gaan over de bediening van de app.
12. Wat vindt u van het bedieningsproces van de app (in hoesje plaatsen, om voet doen, starten, stoppen)?
O
O
O
O
O
Moeilijk/ omslachtig
1 2 3 4 5 Makkelijk/
eenvoudig
13. “De app is gebruiksvriendelijk”
O Helemaal mee oneensO Een beetje oneens O Een beetje eens O Helemaal mee eens
14. “De band om de voet tijdens het uitvoeren van de oefening is niet hinderlijk”
O Helemaal mee oneensO Een beetje oneens O Een beetje eens O Helemaal mee eens
15. Welke verbeterpunten ten aanzien van de bediening kunt u noemen?
………. ………. ……….16. Heeft u nog overige opmerkingen of suggesties?
………. ………. ………. Hartelijk dank voor uw medewerking!
