- Open Access
Validation of a mechanism to balance exercise difficulty in robot-assisted upper-extremity rehabilitation after stroke
© Zimmerli et al; licensee BioMed Central Ltd. 2012
Received: 9 April 2011
Accepted: 3 February 2012
Published: 3 February 2012
The motivation of patients during robot-assisted rehabilitation after neurological disorders that lead to impairments of motor functions is of great importance. Due to the increasing number of patients, increasing medical costs and limited therapeutic resources, clinicians in the future may want patients to practice their movements at home or with reduced supervision during their stay in the clinic. Since people only engage in an activity and are motivated to practice if the outcome matches the effort at which they perform, an augmented feedback application for rehabilitation should take the cognitive and physical deficits of patients into account and incorporate a mechanism that is capable of balancing i.e. adjusting the difficulty of an exercise in an augmented feedback application to the patient's capabilities.
We propose a computational mechanism based on Fitts' Law that balances i.e. adjusts the difficulty of an exercise for upper-extremity rehabilitation. The proposed mechanism was implemented into an augmented feedback application consisting of three difficulty conditions (easy, balanced, hard). The task of the exercise was to reach random targets on the screen from a starting point within a specified time window. The available time was decreased with increasing condition difficulty. Ten subacute stroke patients were recruited to validate the mechanism through a study. Cognitive and motor functions of patients were assessed using the upper extremity section of the Fugl-Meyer Assessment, the modified Ashworth scale as well as the Addenbrookes cognitive examination-revised. Handedness of patients was obtained using the Edinburgh handedness inventory. Patients' performance during the execution of the exercises was measured twice, once for the paretic and once for the non-paretic arm. Results were compared using a two-way ANOVA. Post hoc analysis was performed using a Tukey HSD with a significance level of p < 0.05.
Results show that the mechanism was capable of balancing the difficulty of an exercise to the capabilities of the patients. Medians for both arms show a gradual decrease and significant difference of the number of successful trials with increasing condition difficulty (F 2;60 = 44.623; p < 0.01; η 2 = 0.623) but no significant difference between paretic and non-paretic arm (F 1;60 = 3.768; p = 0.057; η 2 = 0.065). Post hoc analysis revealed that, for both arms, the hard condition significantly differed from the easy condition (p < 0.01). In the non-paretic arm there was an additional significant difference between the balanced and the hard condition (p < 0.01). Reducing the time to reach the target, i.e., increasing the difficulty level, additionally revealed significant differences between conditions for movement speeds (F 2;59 = 6.013; p < 0.01; η 2 = 0.185), without significant differences for hand-closing time (F 2;59 = 2.620; p = 0.082; η 2 = 0.09), reaction time (F 2;59 = 0.978; p = 0.383; η 2 = 0.036) and hand-path ratio (F 2;59 = 0.054; p = 0.947; η 2 = 0.002). The evaluation of a questionnaire further supported the assumption that perceived performance declined with increased effort and increased exercise difficulty leads to frustration.
Our results support that Fitts' Law indeed constitutes a powerful mechanism for task difficulty adaptation and can be incorporated into exercises for upper-extremity rehabilitation.
Motor skills are indispensable to interact with and navigate through our social environment. Neurological disorders such as spinal cord lesions, stroke and traumatic brain injuries frequently affect these motor functions. Since these impairments influence a person's participation in social and economic life, their restoration through rehabilitation plays an important role in improving patients' quality of life .
In recent years the use of robotic interventions have become more and more popular in the field of rehabilitation. Upper- and lower-extremity devices allow patients to participate more actively during therapy, allow longer training periods and more precise repetitions of the motor patterns that have to be regained and assist and reduce the workload of therapists during rehabilitation on a daily basis [2–9]. Nevertheless, patients still rely on physiotherapists to keep them motivated and engaged during therapy and give them feedback about the quality of their movements.
With the advent of modern media technologies, the use of augmented feedback has nowadays become more and more prominent. Augmented feedback applications use digital displays e.g. computer screens and virtual reality (VR) [10, 11] to provide patients with an external feedback source about their motor performance. If correctly applied, augmented feedback applications allow precise movement feedback, supplementing proprioception, and through game-like scenarios increase the overall motivation and engagement during training-key ingredients for productive motor learning and an important factor for successful rehabilitation [1, 12, 13].
Many augmented feedback applications have been developed in the past years. They can generally be divided into applications that increase motivation through improved graphics and challenging exercises as well as applications that concentrate more on direct feedback and lack the motivating aspects of the former [9, 14–24]. Most of these applications are, however, only capable of training a certain aspect of the movement and include a set of predefined exercise difficulty levels. Thus the therapist faces the challenge of deciding on the right application and exercise difficulty that is suitable and most effective for the patient's current stage of recovery.
Due to the increasing number of patients, increasing medical costs and limited therapeutic resources, clinicians may in the future want patients to practice their movements at home or with reduced supervision during their stay in the clinic. The need for augmented feedback applications that are capable of keeping a patient engaged and motivated while respecting a patients' abilities will hence be desirable. Studies have shown that people only engage in an activity if the outcome matches the effort at which they perform [25–27]. Hence learners who believe that they are competent or successful have been shown to remain engaged and motivated over a longer period of time . In our view, this can only be achieved if an augmented feedback application considers the cognitive and physical deficits of patients and incorporates a mechanism that is capable of balancing the difficulty of an exercise i.e. adapt the difficulty to the current capabilities of the patient.
Several studies have already addressed the need for balancing mechanisms in augmented feedback applications for upper-extremity rehabilitation [17, 29, 30] as well as in specifically developed robotic devices [31, 32]. While the proposed approaches in augmented feedback applications have shown to be functional they require the evaluation of the effects that different properties of VR elements have on the level of difficulty, e.g., object size or speed of exercise elements. This can be a complex and enduring process that has to be repeated for each particular application and hence complicates the adaptation to different exercises. Therefore, we propose a different mechanism for upper-extremity rehabilitation that we believe is capable of adapting the difficulty of an exercise. It founds upon a well established empirical formula, Fitts' Law , that natively already incorporates and describes different parameters and their effects on the level of difficulty. Since Fitts' Law is a physiologically valid and widely used descriptor of reaching motions it can be applied to a variety of different upper-extremity exercises. It has successfully been shown that Fitts' Law, which originated from one-dimensional, real-world observations can also be applied to computer screens and input devices [34–36] and is valid in the two dimensional space . The goal of this paper was to apply this mechanism to an augmented feedback application and verify its applicability and validity through a study with ten stroke patients. Since Fitts' law is capable of describing reaching motions, we hypothesize that its application as a balancing mechanism will result in exercise difficulties that are adapted to patients capabilities. We believe that this will lead to a challenging instead of a frustrating therapy experience.
Since upper-extremity rehabilitation of stroke patients is concerned with the reacquisition of reaching motions, Fitts' Law allows to assess the current quality of a patient's movement, thus yielding information about his/her capabilities [38–40]. This information could hence also be used by a mechanism to adjust the difficulty of an exercise.
List of patients that participated in the study
Time since stroke [months]
ACE-R (visuo-spatial subscore)
Weight compensation system
Augmented Feedback Application
Questions of the NASA Task Load Index
How mentally demanding was the task?
How physically demanding was the task
How hurried or rushed was the pace of the task?
How successful were you in accomplishing what you were asked to do?
How hard did you have to work to accomplish your level of performance?
How insecure, discouraged, irritated, stressed and annoyed were you?
Data Acquisition & Analysis
Experiments were conducted on a commercial PC with a screen resolution of 1280 × 1024 pixels using the weight compensation device as mouse input. All statistical analyses and plots were generated using IBM SPSS 19 (IBM Corporation, USA) and MATLAB R2009b (MathWorks, USA) on an Intel MacBook Pro. Parameters were compared using a two-way ANOVA. Post hoc analyses were performed using a Tukey HSD. The significance level was set to p < 0.05.
List of the intercepts, slopes and correlations
In the current study we investigated the applicability and validity of a mechanism that should be useful for adapting the difficulty of the exercise to the capabilities of the patient for upper-extremity rehabilitation. This difficulty balancing mechanism is based on Fitts' Law.
Our results show a decreasing amount of successful trials from the easy to the hard conditions. This seems obvious as patients had less time to reach the target on time and complete the trial. Since the ANOVA did not reveal any significant differences between the paretic and non-paretic arm, we can state that Fitts' Law is a valid mechanism to balance the difficulty of an exercise. Hence, although patients are capable of moving their non-paretic arm more accurately and faster, the number of successful trials is similar compared to when they perform the movements with their paretic arm.
The developed application further allowed assessing different movement characteristics, i.e., movement speed, hand-closing time, reaction time and the hand-path ratio. Unlike for the reaction times all other measurements showed a significant difference between the paretic and non-paretic arm reflecting the impaired motor control. While this seems obvious, the interesting aspect is that movement speed increased and hand-closing time decreased with increasing condition difficulty while the hand-path ratio stayed the same. Although one could assume that an increase of the exercise speed could cause decreased movement accuracy, this was not the case. It rather influenced whether the target was reached or not. Increased speed could however be used to train the hand-closing functionality and movement speed of the paretic arm. A subsequent study should thus assess the additional impact that the balancing mechanism may have on upper-extremity motor recovery after stroke.
Different theories on motivation state that performance is influenced by the effort a person exhibits during an exercise and the outcome they perceive. The effort, performance and frustration sub-scales of the NASA-TLX visualize this. Although the only significant difference for the paretic arm was found in the performance subscale, the others show a trend that, while patients increase their effort with increasing condition difficulty, they perceive their performance as getting worse and hence get frustrated. This eventually would lead to disengagement. Assuming that a success rate of 100% during an exercise my cause boredom and anything below 50% frustration, using Fitts' Law gives an initially challenging but at the same time not frustrating exercise difficulty. By manipulating the estimated time one can adjust the difficulty and thus influence a patients awareness of their capabilities, thereby increasing engagement during therapy.
Despite these findings, some limitations of the current study design need to be noted. In the current setup patients had to complete a trial by closing their hand when they reached the target. While this was useful to gain some insight into patients' hand-closing functionality, it at the same time limited the number of patients that were able to participate. Therefore, in a subsequent study, trial completion could be realized by asking patients to simply move over the target. This would allow inclusion of more patients with the drawback of losing one of the metrics.
In order to acquire the slopes and intercepts for the balancing mechanism, 50 trials were measured during the "assessment phase". As the results of Fitts' Law show, the correlations of the linear regressions are sometimes quite low. This probably caused sporadic inaccurate estimates during the different conditions. According to Figure 5, these, however, did not have a large effect on the number of successful trials. The number of assessment trials was limited to 50 because we did not want patients to get bored by the rather simple task. A subsequent study should assess the minimal amount of trials that are needed in order to obtain sufficiently high correlations.
Another interesting finding when looking at the correlations of the Fitts' Law data is the difference between the paretic and non-paretic arm. While patients 1, 2, 3, 9 and 10 had high correlations (R2 > 0.7) for the paretic arm, they showed low correlations for the non-paretic arm. The opposite was found for patients 4, 5, 6, 7, 8 and 10. Although there seems to be a visible pattern when looking at this distribution, the clinical evaluation does not show any functional differences between the two groups that would allow drawing any conclusions. We rather suppose that the sometimes-low correlations for the non-paretic arm where due to the fact, that patients were not used to train using it. Subsequent studies should however further address this finding.
During the "exercise phase" each condition (easy/balanced/hard) was presented and performance measured for 2 minutes. Therefore the number of trials in each condition differed. This approach, compared to having a fixed number of trials per condition, was chosen to guarantee equal temporal lengths, preventing exhaustion, which could have negatively affected the performance during subsequent conditions.
In order to obtain a subjective measure of condition difficulty, patients were asked to give feedback using the NASA-TLX. Although nearly no significant differences were found, results show a trend indicating that condition difficulty is followed by an increase of mental demand, effort and frustration levels as well as by a decrease of performance. As we, however, also conclude on the motivation level of the patients, using a more established and motivation-oriented questionnaire, e.g., the intrinsic motivation inventory (IMI)  in subsequent studies would be of high importance. In the current study we chose the NASA-TLX over the IMI because of the smaller number of questions patients had to answer. This kept the cognitive demand low during the rest period.
The graphical implementation of the current application was kept quite simple, because the goal of the study was to verify the applicability of the balancing mechanism. We believe that, because of its simplicity, the mechanism can also be adapted to applications with higher graphical details.
We developed an augmented feedback application with a balancing mechanism based on Fitts' Law and validated its applicability through a study with ten stroke patients. We were able to show that Fitts' Law indeed constitutes a powerful mechanism for task difficulty adaptation. Because it is physiologically valid and a widely used descriptor of reaching movements, we believe that Fitts' Law is an ideal estimator for upper-extremity rehabilitation in which patients have to relearn both speed and accuracy of movements. The simplicity of the mechanism further allows incorporating it in a large number of existing and novel augmented feedback applications, independent of the graphical settings those applications exhibit. Since studies have already shown that people only engage in an activity if the outcome matches the effort at which they perform [25–27], we encourage developers to incorporate such balancing mechanisms into their upcoming augmented feedback applications for rehabilitation. This may not only improve the effort that patients exhibit during therapy, but also increase their engagement and motivation over a longer period of time.
This work was supported by the EU Project MIMICS funded by the European Community's Seventh Framework Program (FP7/2007-2013) under grant agreement nr. 215756.
- Holden MK: Virtual environments for motor rehabilitation: review. Cyberpsychol Behav 2005, 8: 187-211. discussion 212-9 10.1089/cpb.2005.8.187View ArticlePubMedGoogle Scholar
- Colombo G, Joerg M, Schreier R, Dietz V: Treadmill training of paraplegic patients using a robotic orthosis. J Rehabil Res Dev 2000, 37: 693-700.PubMedGoogle Scholar
- Colombo G, Wirz M, Dietz V: Driven gait orthosis for improvement of locomotor training in paraplegic patients. Spinal Cord 2001, 39: 252-255. 10.1038/sj.sc.3101154View ArticlePubMedGoogle Scholar
- Housman SJ, Scott KM, Reinkensmeyer DJ: A randomized controlled trial of gravity-supported, computer-enhanced arm exercise for individuals with severe hemiparesis. Neurorehabil Neural Repair 2009, 23: 505-514. 10.1177/1545968308331148View ArticlePubMedGoogle Scholar
- Riener R, Lünenburger L, Jezernik S, Anderschitz M, Colombo G, Dietz V: Patient-cooperative strategies for robot-aided treadmill training: first experimental results. IEEE Trans Neural Syst Rehabil Eng 2005, 13: 380-394. 10.1109/TNSRE.2005.848628View ArticlePubMedGoogle Scholar
- Sanchez R, Reinkensmeyer D, Shah P, Liu J, Rao S, et al.: Monitoring functional arm movement for home-based therapy after stroke. Conf Proc IEEE Eng Med Biol Soc 2004, 7: 4787-4790.PubMedGoogle Scholar
- Kwakkel G, Kollen BJ, Krebs HI: Effects of robot-assisted therapy on upper limb recovery after stroke: a systematic review. Neurorehabil Neural Repair 2008, 22: 111-121.PubMed CentralView ArticlePubMedGoogle Scholar
- Prange GB, Jannink MJ, Groothuis-Oudshoorn CG, Hermens HJ, Ijzerman MJ: Systematic review of the effect of robot-aided therapy on recovery of the hemiparetic arm after stroke. J Rehabil Res Dev 2006, 43: 171-184. 10.1682/JRRD.2005.04.0076View ArticlePubMedGoogle Scholar
- Lünenburger L, Colombo G, Riener R, Dietz V: Biofeedback in gait training with the robotic orthosis Lokomat. Conf Proc IEEE Eng Med Biol Soc 2004, 7: 4888-4891.PubMedGoogle Scholar
- Magill R: Motor learning and control: Concepts and applications. New York: The McGraw-Hill Companies; 2007.Google Scholar
- Linquist B: Motor Learning Concepts and Applications. Pediatric Physical Therapy 2002.Google Scholar
- Fung J, Malouin F, McFadyen BJ, Comeau F, Lamontagne A, et al.: Locomotor rehabilitation in a complex virtual environment. Conf Proc IEEE Eng Med Biol Soc 2004, 7: 4859-4861.PubMedGoogle Scholar
- Keshner EA: Virtual reality and physical rehabilitation: a new toy or a new research and rehabilitation tool? J Neuroeng Rehabil 2004, 1: 8. 10.1186/1743-0003-1-8PubMed CentralView ArticlePubMedGoogle Scholar
- Brütsch K, Schuler T, Koenig A, Zimmerli L, Koeneke SM, et al.: Influence of virtual reality soccer game on walking performance in robotic assisted gait training for children. J Neuroeng Rehabil 2010, 7: 15. 10.1186/1743-0003-7-15PubMed CentralView ArticlePubMedGoogle Scholar
- Cameirao MS, Badia SB, Zimmerli L, Oller ED, Verschure PFMJ: The Rehabilitation Gaming System: a Virtual Reality Based System for the Evaluation and Rehabilitation of Motor Deficits. Virtual Rehabilitation, 2007 2007, 29-33.View ArticleGoogle Scholar
- Cameirao MS, Bermudez IBS, Duarte Oller E, Verschure PF: The rehabilitation gaming system: a review. Stud Health Technol Inform 2009, 145: 65-83.PubMedGoogle Scholar
- Cameirao MS, Badia SB, Oller ED, Verschure PF: Neurorehabilitation using the virtual reality based Rehabilitation Gaming System: methodology, design, psychometrics, usability and validation. J Neuroeng Rehabil 2010, 7: 48. 10.1186/1743-0003-7-48PubMed CentralView ArticlePubMedGoogle Scholar
- Eng K, Siekierka E, Pyk P, Chevrier E, Hauser Y, et al.: Interactive visuo-motor therapy system for stroke rehabilitation. Med Biol Eng Comput 2007, 45: 901-907. 10.1007/s11517-007-0239-1View ArticlePubMedGoogle Scholar
- Koenig A, Brütsch K, Zimmerli L, Guidali M, Duschau-Wicke A, et al.: Virtual environments increase participation of children with cerebral palsy in robot-aided treadmill training. Virtual Rehabilitation, 2008 2008, 121-126.View ArticleGoogle Scholar
- Koenig A, Wellner M, Koneke S, Meyer-Heim A, Lünenburger L, Riener R: Virtual gait training for children with cerebral palsy using the Lokomat gait orthosis. Stud Health Technol Inform 2008, 132: 204-209.PubMedGoogle Scholar
- Piron L, Tonin P, Piccione F, Iaia V, Trivello E, Dam M: Virtual environment training therapy for arm motor rehabilitation. Presence: Teleoperators & Virtual Environments 2005, 14: 732-740. 10.1162/105474605775196580View ArticleGoogle Scholar
- Rizzo A: Expose, distract, motivate and measure: Virtual reality games for health. Nuevas Ideas en Informatico Educativa 2006, 2: 1-4.Google Scholar
- Siekierka EM, Eng K, Bassetti C, Blickenstorfer A, Cameirao MS, et al.: New technologies and concepts for rehabilitation in the acute phase of stroke: a collaborative matrix. Neurodegener Dis 2007, 4: 57-69. 10.1159/000100360View ArticlePubMedGoogle Scholar
- Zimmerli L, Duschau-Wicke A, Riener R, Mayr A, Lünenburger L: Virtual reality and gait rehabilitation Augmented feedback for the Lokomat. Virtual Rehabilitation International Conference, 2009 2009, 150-153.View ArticleGoogle Scholar
- Hill T, Fultz J, Biner PM: Incidental learning as a function of anticipated task difficulty. Motivation and Emotion 1985, 9: 71-85. 10.1007/BF00991551View ArticleGoogle Scholar
- Kukla A: Foundations of an attributional theory of performance. Psychological Review 1972, 79: 454-470.View ArticleGoogle Scholar
- Wright RA: Perceived motivational arousal as a mediator of the magnitude of goal valence. Motivation and Emotion 1982, 6: 161-180. 10.1007/BF00992462View ArticleGoogle Scholar
- Lequerica AH, Kortte K: Therapeutic engagement: a proposed model of engagement in medical rehabilitation. Am J Phys Med Rehabil 2010, 89: 415-422. 10.1097/PHM.0b013e3181d8ceb2View ArticlePubMedGoogle Scholar
- Kan P, Jesse H, Mihailidis A: Automated Upper Extremity Rehabilitation for Stroke Patients Using a Partially Observable Markov Decision Process. AAAI 2008 Fall Symposium on AI in Eldercare 2008.Google Scholar
- Li Y, Huegel JC, Patoglu V, O'Malley MK: Progressive shared control for training in virtual environments. Third Joint Eurohaptics Conference and Symposium on Haptic Interfaces for Virtual Environment and Teleoperation Systems, (WHC'09) 2009, 332-337.Google Scholar
- Colombo R, Pisano F, Mazzone A, Delconte C, Micera S, et al.: Design strategies to improve patient motivation during robot-aided rehabilitation. J Neuroeng Rehabil 2007, 4: 3. 10.1186/1743-0003-4-3PubMed CentralView ArticlePubMedGoogle Scholar
- Krebs HI, Palazzolo JJ, Dipietro L, Ferraro M, Krol J, et al.: Rehabilitation Robotics: Performance-Based Progressive Robot-Assisted Therapy. Auton Robots 2003, 15: 7-20. 10.1023/A:1024494031121View ArticleGoogle Scholar
- Fitts PM: The information capacity of the human motor system in controlling the amplitude of movement. J Exp Psychol 1954, 47: 381-391.View ArticlePubMedGoogle Scholar
- MacKenzie IS, Sellen A, Buxton WAS: A comparison of input devices in element pointing and dragging tasks. Proceedings of the SIGCHI conference on Human factors in computing systems: Reaching through technology 1991, 161-166.View ArticleGoogle Scholar
- MacKenzie IS: Fitts' law as a research and design tool in human-computer interaction. Human-Computer Interaction 1992, 7: 91-139. 10.1207/s15327051hci0701_3View ArticleGoogle Scholar
- Soukoreff RW, MacKenzie IS: Towards a standard for pointing device evaluation, perspectives on 27 years of Fitts' law research in HCI. International Journal of Human-Computer Studies 2004, 61: 751-789. 10.1016/j.ijhcs.2004.09.001View ArticleGoogle Scholar
- MacKenzie IS, Buxton W: Extending Fitts' law to two-dimensional tasks. Proceedings of the SIGCHI conference on Human factors in computing systems 1992, 219-226.Google Scholar
- Christe B, Burkhard PR, Pegna AJ, Mayer E, Hauert CA: Clinical assessment of motor function: a processes oriented instrument based on a speed-accuracy trade-off paradigm. Behavioural Neurology 2007, 18: 19-29.View ArticlePubMedGoogle Scholar
- McCrea PH, Eng JJ: Consequences of increased neuromotor noise for reaching movements in persons with stroke. Experimental Brain Research 2005, 162: 70-77. 10.1007/s00221-004-2106-8PubMed CentralView ArticlePubMedGoogle Scholar
- Smits-Engelsman BCM, Rameckers EAA, Duysens J: Children with congenital spastic hemiplegia obey Fitts' law in a visually guided tapping task. Experimental Brain Research 2007, 177: 431-439. 10.1007/s00221-006-0698-xView ArticlePubMedGoogle Scholar
- Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, Steglind S: The post-stroke hemiplegic patient. 1. a method for evaluation of physical performance. Scand J Rehabil Med 1975, 7: 13-31.PubMedGoogle Scholar
- Bohannon RW, Smith MB: Interrater reliability of a modified Ashworth scale of muscle spasticity. Phys Ther 1987, 67: 206-207.PubMedGoogle Scholar
- Mathuranath PS, Nestor PJ, Berrios GE, Rakowicz W, Hodges JR: A brief cognitive test battery to differentiate Alzheimer's disease and frontotemporal dementia. Neurology 2000, 55: 1613-1620.View ArticlePubMedGoogle Scholar
- Oldfield RC: The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 1971, 9: 97-113. 10.1016/0028-3932(71)90067-4View ArticlePubMedGoogle Scholar
- Hart SG, Staveland LE: Development of NASA-TLX (Task Load Index): Results of empirical and theoretical research. Human mental workload 1988, 1: 139-183.View ArticleGoogle Scholar
- Hart SG: NASA-task load index (NASA-TLX); 20 years later. Human Factors and Ergonomics Society Annual Meeting Proceedings 2006,50(9):904-908. 10.1177/154193120605000909View ArticleGoogle Scholar
- Ryan RM: Control and information in the intrapersonal sphere: An extension of cognitive evaluation theory. Journal of Personality and Social Psychology 1982, 43: 450-461.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.