Maura Casadio Department of Informatics, Systems and Telematics University of Genova Via Opera Pia 13, 16145 Genova GE, Italy and Foundation ‘Don Carlo Gnocchi’ Via Cisa Vecchia, 19038 Sarzana SP, Italy [email protected]

A Haptic Robot Reveals the Adaptation Capability of Individuals with Multiple Sclerosis

Vittorio Sanguineti Department of Informatics, Systems and Telematics and Research Center for Neuroscience and Neuroengineering University of Genova Via Opera Pia 13, 16145 Genova GE, Italy [email protected]

Claudio Solaro Neurology Department, Hospital “P. Antero Micone” Via D. Oliva 22, 16158 Genova GE, Italy [email protected]

Pietro G. Morasso Department of Informatics, Systems and Telematics University of Genova Via Opera Pia 13, 16145 Genova GE, Italy [email protected]

Abstract

1. Introduction

A prerequisite for rehabilitation is that patients preserve their ability to adapt to novel dynamic environments, an ability that has been associated with the cerebellar system. In this study, we use a robot manipulandum to assess the ability of multiple sclerosis (MS) subjects in the early phase of the disease to adapt to a speed-dependent force field. Their performance is compared with an equal number of age-matched controls. We found that MS subjects display subtle incoordination problems but do not significantly differ from controls in their ability to adapt to the force field. These findings are discussed in terms of the possible benefits that MS subjects might receive from robot-assisted therapy that is specifically aimed at impaired visuomotor coordination.

KEY WORDS—rehabilitation robotics, motor learning, multiple sclerosis The International Journal of Robotics Research Vol. 26, No. 11–12, November/December 2007, pp. 1225–1233 DOI: 10.1177/0278364907084981 c SAGE Publications 2007 Los Angeles, London, New Delhi and Singapore 1

Over the last 20 years, robots have been used widely in the experimental investigation of the mechanisms underlying the neural control of movement. In a typical application, robots generate controlled perturbations to ongoing movements. This allows the response of the motor system to such perturbations, as well as the underlying control modalities, to be quantified (Mussa-Ivaldi et al. 19851 Gomi and Kawato 1996). Robotic technology has also shown promise recently in using specialized forces that stimulate adaptation in the nervous system as a mode of recovery. The adaptive properties of the motor system have been studied in experiments in which robots deliver forces that may be made dependent on position, speed and/or acceleration, thus allowing specific dynamic environments (or “force fields”) to be simulated. Most of these studies involved planar arm movements (Shadmehr and Mussa-Ivaldi 1994), but applications to orofacial control have been reported (Tremblay et al. 2003). Adaptation to force fields may be seen as a generalization of 1225

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008

1226

THE INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH / November/December 2007

perturbation studies. In fact, the patterns of sensorimotor adaptation to unfamiliar dynamic environments may indirectly provide information about the body’s mechanical impedance at a specific activation, configuration and speed (Scheidt et al. 20011 Takahashi et al. 20011 Donchin et al. 2003). Adaptation studies have provided a large body of knowledge on the mechanisms underlying the way the brain reacts to novel dynamic environments (Shadmehr and Wise 2005). Robots are being recognized as promising tools for the rehabilitation of upper-limb impairment. They may be used to guide or assist the movements of a patient in close interaction, in much the same way as a human physical therapist. Clinical trials (Prange et al. 2006) suggest that robot therapy is quite effective: for instance, the MIT-MANUS system (Krebs et al. 1999) was demonstrated to accelerate the recovery of stroke patients (Volpe et al. 1999). Most applications of robot therapy involve acute or chronic stroke patients and often mimic the exercises delivered by a human therapist. However, robot therapy would be even more appealing in the treatment of deficits for which no rehabilitation methods have proved effective, such as those involving the coordination of movements with multiple degrees of freedom, caused by lesions to the cerebellum and/or the basal ganglia. A prerequisite for rehabilitation, whether robot- or therapist-assisted, is that patients preserve their ability to adapt to novel dynamic environments, an ability related to the feedforward component of control. Recently, an experimental protocol that is widely used to investigate motor learning capabilities (Shadmehr and Mussa-Ivaldi 1994) was applied to subjects with degenerative cerebellar atrophy (Maschke et al. 20041 Smith and Shadmehr 2005) and it was shown that these subjects completely lost their ability to adapt. In contrast, this same ability is preserved in subjects with Huntington’s disease (Smith and Shadmehr 2005) as well as in stroke survivors (Takahashi and Reinkensmeyer 20031 Patton et al. 2006). Multiple sclerosis (MS) is the second biggest cause of neurological disability among young adults. Therefore, it seems natural to explore the potential of robots in the assessment and treatment of these patients. MS is usually described as a chronic autoimmune disease, characterized by inflammatory demyelination, affecting the white matter in the central nervous system. This results in the impairment of multiple functional systems, in proportions that change widely from patient to patient. Common symptoms include weakness of one or more extremities, muscle spasticity, double vision, urinary incontinence and loss of coordination. About 85% of the subjects show a relapsing–remitting course, i.e. acute phases (“relapses”) alternating with intervals in which the patient’s conditions remain stable. In the early stages of MS, although axonal injury is not reversible, relapses are often followed by partial or complete functional recovery (“remissions”). Clinical recovery is determined by many factors, such as increased expression of sodium channels, recruitment of silent pathways and remyelination. Cortical reorganization may take place as

Fig. 1. Experimental set-up.

well (Rocca et al. 2003), thus playing a role in limiting the impact of axonal loss. Reorganization may leave performance relatively unaffected, but the compensatory strategies that are exploited in controlling the movements may be unveiled by highly sensitive experimental and analytic techniques, aimed at investigating sensorimotor control and adaptation. In this paper, we describe an application of the abovementioned adaptation protocol to the assessment of sensorimotor performance and adaptation in subjects with a confirmed diagnosis of MS. As functional brain imaging studies have suggested that MS subjects with no disability display a substantial cortical and sub-cortical reorganization (Rocca et al. 2005), we asked whether MS subjects with sub-clinical sensorimotor symptoms display signs of compensatory strategies, and whether and to what extent these subjects preserve their ability to reorganize their sensorimotor behavior in order to adapt to an unfamiliar dynamical environment. We also discuss the diagnostic implications and the perspectives for robot-assisted rehabilitation with these patients and MS patients in general.

2. The Robotic Manipulandum The robotic manipulandum that we used in this study (see Figure 1) was specifically designed for the evaluation of motor learning and control, and for robot therapy.

2.1. Robot Architecture The manipulandum (details are reported in Casadio et al. (2006)) has a planar, 80 2 40 cm elliptic workspace that can be rotated around a horizontal axis in order to work on nonhorizontal planes (this feature was not used in the reported experiments). The geometry of the robot was specified as the result of the optimization of a global isotropy index (Stocco et al. 1998).

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008

Casadio, Sanguineti, Solaro, and Morasso/ A Haptic Robot Reveals the Adaptation Capability of Individuals with MS The robot is actuated by two direct-drive brushless motors, mounted proximally to minimize the overall inertia at the handle (less than 1 kg) as well as the frictional forces (less than 0.06 N)1 the manipulability index and the force/torque ratio are quite uniform over the whole workspace (23 3 2 cm and 212 3 012 N Nm41 , respectively). The force available at the handle is greater than 50 N (continuous) and greater than 200 N (peak), in all directions. Rotations of the motors were estimated by a pair of sin/cos encoders (equivalent resolution: 17 bits), which allowed a spatial resolution of less than 0.1 mm over the whole workspace. We did not use a force sensor at the handle, because the extremely low friction and the low inertia allow to the interaction forces from the motor currents to be estimated directly. Although the robot is in many respects similar to the MITMANUS system, there are important differences in the mechanical design (the ability to rotate the plane of movements) and in the available power (steady hand forces are almost three times greater). 2.2. Control Architecture The control architecture consists of an inner current loop (running at 16 kHz) and an outer impedance control loop (at 1 kHz). The current loop is implemented by two control units (one for each motor). The impedance control loop runs on a dedicated personal computer, under a real-time operating system (QNX). An additional computer is used for configuration and command and also runs the graphical user interface. The impedance control scheme is defined by the following equation: 7 x34 8 (1) Tm 5 J 2q3T 6 Fh 2x4 x4 where Tm is the torque vector to be generated by the two motors, J 2q3 is the Jacobian matrix of the manipulator and q is the vector of joint angles. In general, the specified impedance function at the handle, Fh , is a function of the position of the handle (x) and of its time derivatives. The software environment for the control and design of the exercise protocols is based on a Simulink fast-prototyping enR 1 vironment, RT-Lab (Opal-RT Technologies Inc.). For a given exercise, a set of visual objects is specified and displayed on a computer screen. Visual objects are represented in terms of the Virtual Reality Modeling Language (VRML), by using Simulink’s Virtual Reality toolset. The exercise protocol specifies the interaction between the robot and visual objects. This is specified as a finite-state machine, implemented by R 1 means of a standard Matlab tool, Stateflow .

3. Quantifying Sensorimotor Performance 3.1. Experimental Protocol and Task In experiments aimed at the assessment of sensorimotor performance, the robot is used as a passive device. Motors are

1227

turned off and the encoders are used to record hand trajectories. Subjects sit on a chair, with their torso and wrist restrained by means of suitable holders, and grasp the handle of the manipulandum with their dominant hand. The forearm is supported by a low-friction sled on the horizontal surface of a table. The height of the seat is adjusted so that the arm can be kept horizontal at the level of the shoulder joint. Therefore, only the shoulder and elbow could move and motion was restricted to the horizontal plane, with no influence of gravity. The position of the seat is also adjusted in such a way that when the cursor is pointing at the center of the workspace, the elbow and the shoulder joints are flexed about 909 and 459 , respectively. The task consists of making movements in eight different directions, starting from the same initial position at the center of the workspace. The targets were presented on a 19” LCD computer screen, placed in front of the subjects, about 1 m away, at eye level. Targets are displayed as round green circles (2 cm diameter) against a black background. The current position of the hand in the workspace is also continuously displayed as a yellow circle (0.4 cm diameter). The amplitude of the movements (distance of the targets from the center) was 10 cm. The sequence of target presentations alternated the central target and one of four peripheral targets, generated in random order. In order to decrease movement variability, the subjects were encouraged to keep an approximately constant timing. We set the desired duration to 500 3 50 ms. If the estimated duration was inside this range, a positive feedback/reward to the subject (a pleasant sound) was provided. If the measure was below or above that range, no sound was generated and the color of the target was changed to red or white, respectively. We also informed the subjects that the reaction time was not important—they could wait as long as they wanted after target appearance before starting each movement—but when ready, they had to perform a single, rapid movement toward the target. The experiment was organized into target sets, each consisting of a sequence of target presentations in which each peripheral target occurred 12 times, for a total 12 2 4 5 48 center– out movements (corresponding to the directions: 09 , 459 , 909 and 1359 ), plus the corresponding 48 return movements (1809 , 2259 , 2709 and 3159 ). The endpoint of each movement was used as the starting point for the subsequent movement. 3.2. Data Analysis Hand trajectories were sampled at 100 Hz. The x and y components were smoothed with a sixth-order Savitzky–Golay filter (window size 270 ms, equivalent cut-off frequency of around 7 Hz), which also allowed... us to estimate the first three time ... derivatives (x4 7 x4 8 x 4 y7 4 y8 4 y ). We then estimated the following indicators.

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008

1228

THE INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH / November/December 2007

Movement Duration

Jerk Ratio

This is the time elapsed between movement onset and termination. Movement onset is computed as the first time instant that the hand speed exceeds a threshold equal to 12% of the peak velocity, at least for 200 ms. Movement termination is computed as the first time instant after onset when movement speed goes below the threshold and stays there for at least 200 ms.

This is the ratio of the jerk indexes (see the previous section) calculated during the deceleration (after the peak in the speed profile) and acceleration phase of the movement. It indicates a difficulty in compensating for self-generated errors and therefore an abnormal ability to exploit sensory information, as in this case trajectories would be expected to be less smooth toward their end (Smith et al. 2000).

Linearity

4. Quantifying Sensorimotor Adaptation

This is the percentage increment of the length of the trajectory traced by the hand, between the onset and termination times, with respect to the straight line (the distance between the initial and final points of the trajectory). It is a measure of path curvature.

Aiming Error This is the difference between the target direction and the actual movement direction in the early phase of the movement (until 100 and 300 ms after movement onset). The 100 ms aiming error is indicative of the performance of the feed-forward component of control. In contrast, the 300 ms error is a general measure of curvature, because at this time of the movement the lateral deviation is largest (Smith et al. 20001 Smith and Shadmehr 2005).

Symmetry This is the ratio between the duration of the acceleration and deceleration phases. If the speed profile is bell-shaped, with equal durations of the acceleration and deceleration phase, symmetry is around 1.

Jerk Index This is the square root of the jerk (norm of the third time derivative of the trajectory), averaged over the overall movement duration and normalized with respect to duration and path length (Teulings et al. 1997): 123 4 T5 1

J 2t3 2 dt 6 2 4 (2) jerk index 5 2 L where T is the movement duration and L is the path length. It is a measure of smoothness: large jerk indexes correspond to less smoothness. The square root is used to compress the large range of variation of the jerk integral.

4.1. Experimental Protocol and Task In adaptation experiments, we used the well-known forcefield adaptation paradigm (Shadmehr and Mussa-Ivaldi 1994), which allows the study of both unperturbed reaching movements and the modifications induced by an unfamiliar artificial dynamic environment. The experimental protocol was organized into three phases: (i) null field, in which the robot generates no force (five target sets)1 (ii) force field, in which the force field was turned on (five target sets)1 and (iii) the after effect (two target sets). Each set lasted approximately 5 min and the subjects were allowed to rest between sets. The purpose of the null-field phase was to set a background level of performance. During the force-field phase, the manipulandum generated a viscous curl field (Shadmehr and MussaIvaldi 1994), i.e. a force field that perturbs the movements by generating forces that are perpendicular to the instantaneous velocity vector of the hand and have a magnitude proportional to the speed. In the impedance control framework, this can be achieved by using the following control equation: 5 F 56

0

4B

B

0

7 8 6 x4 7

(3)

where B is a viscous coefficient that we set to B 5 13 N m41 s41 . This value corresponds to peak forces of 4– 6 N. The hand velocity vector x7 was estimated online as follows: (1) read the joint angles provided by the high-precision encoders1 (2) compute the hand position by means of the direct kinematic equations1 (3) estimate hand velocity by means of a numerical differentiation technique. During the field sets, we randomly inserted “catch trials”, i.e. reaching movements in which the force field was unexpectedly turned off. The purpose of these trials is to sample the progress of the internal representation that is supposedly learnt by the subjects during field adaptation. In our experiments the probability of catch trials was one in six, corresponding to one catch trial per direction per trial set.

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008

Casadio, Sanguineti, Solaro, and Morasso/ A Haptic Robot Reveals the Adaptation Capability of Individuals with MS

1229

4.2. Data Analysis The nervous system may react to the perturbations introduced by the curl field in different ways: (1) it may just ignore it, and accept the distortion introduced by the field, a viable alternative if we consider that this kind of perturbation does not specifically affect the achievement of the target1 (2) it may resist the perturbation by increasing the joint stiffness (a strategy of coactivation), without changing the underlying motor commands1 (3) it may compensate for the perturbation by means of a suitable internal model that learns to modify the background reaching patterns by producing appropriate feed-forward commands. The first alternative can be ruled out if we can demonstrate that the response patterns to the curl field do change during the experimental sessions, a finding that is well established for normal subjects but needs to be demonstrated for the MS patients. The second alternative can be ruled out if we observe the catch trials. In fact, if the reduction in the number of errors following the introduction of a force field is due to an increase in the joint stiffness alone, we should only observe a reduced scatter and an improved straightness of the “catch trial” paths (Burdet et al. 2006). In contrast, if error reduction is a consequence of learned feed-forward control, the unexpected disappearance of the field in the catch trials should generate errors in the opposite direction. The problem is to define a learning index, which should describe the learning process in a quantitative way independently of the magnitude of the force field and user-specific parameters such as the net compliance of the arm. The solution that we used in this study is similar to that proposed by Criscimagna-Hemminger et al. (2003) and used by Smith and Shadmehr (2005): learning index 5

yc yc 4 yf 

(4)

where yf and yc are the 300 ms aiming errors in the field trials and catch trials, respectively. Both error measures were adjusted for any bias that may have been present during the last null-field set. Therefore, errors in a field set always refer to changes from errors in the null set.

5. Experimental Results Eleven subjects with clinically definite, relapsing–remitting MS, according to Poser criteria (Poser et al. 1983), participated in this study. The inclusion criteria were: (i) Extended Disability Status Scale (EDSS) at most 1 (the presence of only neurological signs, but no sign or symptoms at upper limbs)1 (ii) “normal” score for the “arm” portion of the Scripps Neurological Rating Scale (NRS) (Sipe et al. 1984) for the sensory, motor and cerebellar systems. The exclusion criteria were: (i) relapses within the last three months1 (ii) treatment with corticosteroids within the previous three months1 (iii) Mini Mental

Fig. 2. Trajectories in the different phases of the experiments, for typical control (left) and MS subjects (right).

State Examination (MMSE) less than 24. This would correspond to below-normal cognitive abilities. The performance of these subjects was compared with 11 age-matched controls. A sample of the movements recorded during the various experimental phases (null field, force field and after effect) is displayed in Figure 2 for typical control and MS subjects. This figure suggests that the performance of MS subjects differs from that of controls. Based on the above indicators, we compared the performance of MS subjects with that of controls.

5.1. Sensorimotor Performance To test whether and how the motor performance of MS subjects differs from that of controls, we took the null-field portion of the adaptation experiment, when subjects had to become familiarized with the robotic device. More specifically, we looked at the effects of disease (control, MS) in the motor performance, quantified through the indicators defined above, and at how disease affects their time course during the nullfield phase. We found that in MS subjects, movements lasted longer (0159 3 0102 s for controls versus 0177 3 0104 s for MS subjects, mean 3 SE1 p 5 0100027, two-way repeatedmeasures analysis of variance (ANOVA)) and their speed profile was significantly less symmetric with respect to controls (0179 3 0103 versus 0170 3 01031 p 5 01041). Moreover, in MS subjects, movements are more curved (418 3 014 % vs 918 3 213 %1 p 5 01046) and less smooth (714 3 015 versus 1111 3 0171 p 5 0100054) than those of controls. Finally, MS

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008

1230

THE INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH / November/December 2007

subjects revealed a significantly greater 100 ms aiming error (1114 3 0159 versus 1515 3 0169 1 p 5 0100003). During the familiarization phase, both control and MS subjects gradually improved their performance, as reflected by the fact that all indicators displayed a significant time effect. We asked whether there are differences in the rate of improvement in MS subjects and controls (interaction between disease and time). However, none of the above-mentioned indicators displayed significant disease–time interactions.

5.2. Force-1eld Adaptation We then focused on force-field adaptation and asked whether MS subjects could learn to compensate for the force field. For each performance indicator, we ran a repeated-measures ANOVA with two factors: disease and time (early, late) over the force field target sets. We found significant effects ( p 5 0105) of disease (all indicators excluding the jerk ratio) and highly significant ( p 5 0101) effects of time (all indicators excluding the 100 ms aiming error). The lack of significance of disease in the jerk ratio suggests that under the effect of the force field, although their sensorimotor performance is distinctively different, MS and control subjects have similar performance in making feedback corrections. The lack of time effect in the 100 ms aiming error is due to the fact that the speed-dependent field does not affect the early portion of the movement, in which the speed is low1 see Figure 3. As regards the interactions between disease and time during field adaptation, only the jerk index displayed a significant effect ( p 5 01029). A closer look at Figure 3 suggests that this is due to the fact that MS subjects have a greater improvement. In conclusion, MS subjects and controls are basically similar in their ability to adapt and in their rate of adaptation. This is clearly shown in Figure 3 for all indicators. We finally asked whether MS subjects differ from controls in their rate of adaptation. To do this, we compared the learning index (Equation (4)) estimated from MS and control subjects. Even in this case, we found no significant differences. This is clearly shown in Figure 4.

6. Discussion 6.1. Abnormal Sensorimotor Performance and Preserved Adaptation Capability Analysis of the motor performance during the null-field phase suggests that MS patients who are asymptomatic on clinical examination indeed have subtle incoordination problems. Their movements are more curved, less smooth and have a greater aiming error, i.e. they start in the wrong direction. This is consistent with previous findings (Solaro et al. 2007), in

which hand trajectories were recorded with a digitizing tablet. In the present experiment, the observed abnormalities may be enhanced by the non-neglectable dynamics of the manipulandum. The greater aiming error suggests that MS subjects generate inappropriate motor commands. This may reflect an inaccurate account for the anisotropy of the inertia of the arm and the robot system. Normal subjects make errors as well (Gordon et al. 1994), but MS subjects display an error of greater magnitude. Nevertheless, such errors do not prevent subjects from reaching the targets. This is likely because these errors are compensated for, at least in part, by making online corrections based on the available visual and/or proprioceptive information. This may explain the increased magnitude of the jerk index. As regards the ability to adapt to the force field, we found no significant differences between MS subjects and controls in the actual ability to adapt and in the magnitude and rate of adaptation. Therefore, like controls, MS subjects effectively learn to predict the force field generated by the robot rather than just trying to resist to perturbations by stiffening their arm. It should be noted, however, that the failure to detect a difference between groups does not rule out the possibility that a more sensitive test might actually detect a difference in the way adaptation is achieved. 6.2. Fatigue in MS Patients One major concern with MS subjects is that they become easily fatigued. In the present experiments, subjects were allowed to rest between consecutive target sets1 however, only one MS subject actually did. This suggests that the task was well tolerated. Moreover, Figure 3 and statistical analysis shows that there was no degradation of performance at the end of the recovery phase as compared with the final portion of the null phase. This, again, suggests that subjects displayed no problems in performing the task in the late phase of the experiment. Indeed, a small improvement was often observed. 6.3. Implications for Robot Therapy in MS Subjects What are the implications of these results for rehabilitation, considering that no consensus has been reached so far on the most effective approaches to robot therapy? The proposed approaches fall into three basic categories, i.e. assistive, perturbing and adaptive. In assistive approaches, the robot provides assistive or restoring forces, which may vary in proportion to the instantaneous difference between desired and actual movement (Krebs et al. 1998) or on the basis of a position-dependent force field, directed to the intended target (Casadio et al. 2006). In perturbing approaches, the robot generates disturbances while the subject is performing a movement or maintaining a

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008

Casadio, Sanguineti, Solaro, and Morasso/ A Haptic Robot Reveals the Adaptation Capability of Individuals with MS

1231

Fig. 3. Time course of movement performance indicators during the different phases of the experiment (N: null field, F: force field). Thin curves indicate 3 SE. particular posture. The notion that perturbations promote recovery is supported by clinical and physiological observations (Takahashi et al. 20011 Patton et al. 2006). Adaptive approaches rely on observations that are similar to those reported here: exposition to a force field induces an adaptation, which results in a gradual recovery of straight movements. This occurs by building an internal representa-

tion of dynamics, as demonstrated by the presence of “after effects” when the field is removed. Such force fields may be customized to the individual subject (Patton and MussaIvaldi 2004). However, the very same motor adaptation protocol described here may also be used as an adaptive therapy (Patton et al. 2006). In chronic stroke survivors, forcefield training resulted in an improvement of motor perfor-

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008

1232

THE INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH / November/December 2007 and external environment would seem appropriate for these patients. 6.4. Future of Robotics Technology in the Rehabilitation of MS Subjects

Fig. 4. Top: aiming error (300 ms) for control and MS subjects. The dashed curves refer to the catch trials. Bottom: learning index estimated from the indicator above.

mance during the late after-effect phase with respect to the baseline. In our experiments, impairment is too small to allow any observation of improvement of this kind, but this same protocol might be beneficial for MS patients with a more severe impairment. In the case of cerebellar lesions, there is some indication (Bastian et al. 19961 Sanguineti et al. 2003) that defective coordination may depend on an abnormal capability of anticipating the muscle activations needed to compensate for arm dynamics (feed-forward control). The associated symptoms include ataxia (an inability to organize coordinated movements that involve multiple joints) and kinetic tremor. At present, no effective treatments are available for this condition, which is extremely disabling and is common in patients with MS: impairment of the cerebellar system has been estimated to be present in about 80% of MS patients. Therefore, to answer the question that we formulated at the beginning of the section, we suggest that techniques that specifically train the ability to predict the dynamics of body

Even though the majority of studies regarding the evidence for efficacy of rehabilitation in MS are on patients with chronic progressive MS (Freeman et al. 1997), there is growing evidence that patients with MS can benefit from rehabilitation interventions after an acute relapse with incomplete recovery (Liu et al. 2003). Some studies suggest that some cortical reorganization in patients with MS may occur, but it is unclear whether this plays a role in MS rehabilitation (Rasova et al. 2005). The main effect is likely to be a consequence of improved compensation, adaptation and reconditioning (Rocca et al. 2002). The above considerations suggest that in the rehabilitation of MS subjects, robots should primarily aim at promoting the development and the maintenance of compensatory and/or adaptive capabilities. Moreover, owing to the peculiar variability of types and degrees of impairment found in MS subjects, the timing and mode of rehabilitation treatment in MS patients should be set individually, taking into account the degree and extent of impairment. This points to the need for integration between rehabilitation exercises and the continuous monitoring of the control and adaptation performance. Robots should be able to discriminate between the ways in which different pathological conditions can affect the control patterns and/or the learning capability. In particular, it makes sense to independently assess the degree of impairment of the two aspects (control and learning) in the early phases of the disease in which the functional impairment is still mild and results in little or no disability, and there is ground for very focused interventions. This study suggests that robotic technology is potentially useful in both types of assessment. It remains to be seen in further research whether robotic devices may be effective in restoring movement ability in these patients.

References Bastian, A. J. et al. (1996). Cerebellar ataxia: abnormal control of interaction torques across multiple joints. Journal of Neurophysiology, 76(1): 492–509. Burdet, E. et al. (2006). Stability and motor adaptation in human arm movements. Biological Cybernetics, 94(1): 20– 32. Casadio, M. et al. (2006). Impedance controlled, minimally assistive robotic training of severely impaired hemiparetic patients. Proceedings of the 1st IEEE/RAS-EMBS International Conference on Biomedical Robotics and Biomechatronics, Pisa, Italy.

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008

Casadio, Sanguineti, Solaro, and Morasso/ A Haptic Robot Reveals the Adaptation Capability of Individuals with MS Casadio, M. et al. (2006). Braccio di Ferro: a new haptic workstation for neuromotor rehabilitation. Technology and Health Care, 13: 1–20. Criscimagna-Hemminger, S. E. et al. (2003). Learned dynamics of reaching movements generalize from dominant to nondominant arm. Journal of Neurophysiology, 89(1): 168– 176. Donchin, O. et al. (2003). Quantifying generalization from trial-by-trial behavior of adaptive systems that learn with basis functions: theory and experiments in human motor control. Journal of Neuroscience, 23(27): 9032–9045. Freeman, J. A. et al. (1997). The impact of inpatient rehabilitation on progressive multiple sclerosis. Annals of Neurology, 42(2): 236–244. Gomi, H. and Kawato, M. (1996). Equilibrium-point control hypothesis examined by measured arm stiffness during multijoint movement. Science, 272(5258): 117–120. Gordon, J. et al. (1994). Accuracy of planar reaching movements. II. Systematic extent errors resulting from inertial anisotropy. Experimental Brain Research, 99(1): 112–130. Krebs, H. I. et al. (1998). Robot-aided neurorehabilitation. IEEE Transactions on Rehabilitation Engineering, 6(1): 75–87. Krebs, H. I. et al. (1999). Overview of clinical trials with MITMANUS: a robot-aided neuro-rehabilitation facility. Technology and Health Care, 7(6): 419–423. Liu, C. et al. (2003). Does neurorehabilitation have a role in relapsing–remitting multiple sclerosis? Journal of Neurology, 250(10): 1214–1218. Maschke, M. et al. (2004). Hereditary cerebellar ataxia progressively impairs force adaptation during goal-directed arm movements. Journal of Neurophysiology, 91(1): 230– 238. Mussa-Ivaldi, F. A. et al. (1985). Neural, mechanical, and geometric factors subserving arm posture in humans. Journal of Neuroscience, 5(10): 2732–2743. Patton, J. L. and Mussa-Ivaldi, F. A. (2004). Robotassisted adaptive training: custom force fields for teaching movement patterns. IEEE Transactions on Biomedical Engineering, 51(4): 636–646. Patton, J. L. et al. (2006). Evaluation of robotic training forces that either enhance or reduce error in chronic hemiparetic stroke survivors. Experimental Brain Research, 168(3): 368–383. Poser, C. M. et al. (1983). New diagnostic criteria for multiple sclerosis: guidelines for research protocols. Annals of Neurology, 13(3): 227–231 Prange, G. B. et al. (2006). Systematic review of the effect of robot-aided therapy on recovery of the hemiparetic arm after stroke. Journal of Rehabilitation Research and Development, 43(2): 171–184. Rasova, K. et al. (2005). Is it possible to actively and purposely make use of plasticity and adaptability in the neuroreha-

1233

bilitation treatment of multiple sclerosis patients? A pilot project. Clinical Rehabilitation, 19(2): 170–181. Rocca, M. A. et al. (2002). Adaptive functional changes in the cerebral cortex of patients with nondisabling multiple sclerosis correlate with the extent of brain structural damage. Annals of Neurology, 51(3): 330–339. Rocca, M. A. et al. (2003). Functional cortical changes in patients with multiple sclerosis and nonspecific findings on conventional magnetic resonance imaging scans of the brain. Neuroimage, 19(3): 826–836. Rocca, M. A. et al. (2005). Cortical adaptation in patients with MS: a cross-sectional functional MRI study of disease phenotypes. The Lancet Neurology, 4(10): 618–626. Sanguineti, V. et al. (2003). Cerebellar ataxia: quantitative assessment and cybernetic interpretation. Human Movement Science, 22(2): 189–205. Scheidt, R. A. et al. (2001). Learning to move amid uncertainty. Journal of Neurophysiology, 86(2): 971–985. Shadmehr, R. and Mussa-Ivaldi, F. A. (1994). Adaptive representation of dynamics during learning of a motor task. Journal of Neuroscience, 14(5 Pt 2): 3208–3224. Shadmehr, R. and Wise, S. P. (2005). The Computational Neurobiology of Reaching and Pointing: a Foundation for Motor Learning. Cambridge, MA, MIT Press. Sipe, J. C. et al. (1984). A neurologic rating scale (NRS) for use in multiple sclerosis. Neurology, 34(10): 1368–1372. Smith, M. A. et al. (2000). Motor disorder in Huntington’s disease begins as a dysfunction in error feedback control. Nature, 403(6769): 544–549. Smith, M. A. and Shadmehr, R. (2005). Intact ability to learn internal models of arm dynamics in Huntington’s disease but not cerebellar degeneration. Journal of Neurophysiology, 93(5): 2809–2821. Solaro, C. et al. (2007). Subtle upper limb impairment in asymptomatic multiple sclerosis subjects. Multiple Sclerosis 13(3): 428–432. Stocco, L. et al. (1998). Past constrained global minimax optimization of robot parameters. Robotica, 16: 595–605. Takahashi, C. D. and Reinkensmeyer, D. J. (2003). Hemiparetic stroke impairs anticipatory control of arm movement. Experimental Brain Research, 149(2): 131–140. Takahashi, C. D. et al. (2001). Impedance control and internal model formation when reaching in a randomly varying dynamical environment. Journal of Neurophysiology, 86(2): 1047–1051. Teulings, H. L. et al. (1997). Parkinsonism reduces coordination of finger, wrist, and arm fine motor control. Experimental Neurology, 146: 159–170. Tremblay, S. et al. (2003). Somatosensory basis of speech production. Nature 423(6942): 866–869. Volpe, B. T. et al. (1999). Robot training enhanced motor outcome in patients with stroke maintained over 3 years. Neurology 53(8): 1874–1876.

Downloaded from http://ijr.sagepub.com by Pietro Morasso on October 18, 2008