Fully implanted adaptive deep brain stimulation in freely moving essential tremor patients.

OBJECTIVE: Deep brain stimulation (DBS) is a safe and established treatment for essential tremor (ET) and several other movement disorders. One approach to improving DBS therapy is adaptive DBS (aDBS), in which stimulation parameters are modulated in real time based on biofeedback from either external or implanted sensors. Previously tested systems have fallen short of translational applicability due to the requirement for patients to continuously wear the necessary sensors or processing devices, as well as privacy and security concerns. APPROACH: We designed and implemented a translation-ready training data collection system for fully implanted aDBS. Two patients chronically implanted with electrocorticography strips over the hand portion of M1 and DBS probes in the ipsilateral ventral intermediate nucleus of the thalamus for treatment of ET were recruited for this study. Training was conducted using a translation-ready distributed training procedure, allowing a substantially higher degree of control over data collection than previous works. A linear classifier was trained using this system, biased towards activating stimulation in accordance with clinical considerations. MAIN RESULTS: The clinically relevant average false negative rate, defined as fraction of time during which stimulation dropped below [Formula: see text] clinical levels during movement epochs, was 0.036. Tremor suppression, calculated through analysis of gyroscope data, was 33.2% more effective on average with aDBS than with continuous DBS. During a period of free movement with aDBS, one patient reported a slight paresthesia; patients noticed no difference in treatment efficacy between systems. SIGNIFICANCE: Here is presented the first translation-ready training procedure for a fully embedded aDBS control system for MDs and one of the first examples of such a system in ET, adding to the consensus that fully implanted aDBS systems are sufficiently mature for broader deployment in treatment of movement disorders.

Real-time gait event detection for paraplegic FES walking.

A real-time method for the detection of gait events that occur during the electrically stimulated locomotion of paraplegic subjects is described. It consists of a two-level algorithm for the processing of sensor signals and the determination of gait event times. Sensor signals and information about the progression of the stimulator though its pre-specified stimulation "pattern" are processed by a machine intelligence (fuzzy logic) algorithm to determine an initial estimate of the patient's current phase of gait. This is then reviewed and modified by a second algorithm that removes spurious gait estimates, and determines gait event times. These gait event times are known to the system within approximately one-half of a gait cycle. The resulting gait event detection system was successfully evaluated on three subjects. Detection accuracy is not adversely affected by day-to-day gait variability. This work resolved technical and practical issues that previously limited the real time application of these methods. In particular, cosmetically acceptable insole force transducers were used. This gait event detector is designed for use in a real time controller for the automatic adjustment of the intensity and timing of stimulation while the subject is walking using functional electrical stimulation (FES).

Identification of electrically stimulated quadriceps muscles in paraplegic subjects.

This work establishes a method for the noninvasive in vivo identification of parametric models of electrically stimulated muscle in paralyzed individuals, when significant inertial loads and/or load transitions are present. The method used differs from earlier work, in that both the pulse width and stimulus period (interpulse interval) modulation are considered. A Hill-type time series model, in which the output is the product of two factors (activation and torque-angle) is used. In this coupled model, the activation dynamics depend upon velocity. Sequential nonlinear least squares methods are used in the parameter identification. The ability of the model, using identified time-varying parameters, to accurately predict muscle torque outputs is evaluated, along with the variability of the identified parameters. This technique can be used to determine muscle parameter models for biomechanical computer simulations, and for real-time adaptive control and monitoring of muscle response variations such as fatigue.

Using evoked EMG as a synthetic force sensor of isometric electrically stimulated muscle.

A method for the estimation of the force generated by electrically stimulated muscle during isometric contraction is developed here. It is based upon measurements of the evoked electromyogram (EMG) [EEMG] signal. Muscle stimulation is provided to the quadriceps muscle of a paralyzed human subject using percutaneous intramuscular electrodes, and EEMG signals are collected using surface electrodes. Through the use of novel signal acquisition and processing techniques, as well as a mathematical model that reflects both the excitation and activation phenomena involved in isometric muscle force generation, accurate prediction of stimulated muscle forces is obtained for large time horizons. This approach yields synthetic muscle force estimates for both unfatigued and fatigued states of the stimulated muscle. In addition, a method is developed that accomplishes automatic recalibration of the model to account for day-to-day changes in pickup electrode mounting as well as other factors contributing to EEMG gain variations. It is demonstrated that the use of the measured EEMG as the input to a predictive model of muscle torque generation is superior to the use of the electrical stimulation signal as the model input. This is because the measured EEMG signal captures all of the neural excitation, whereas stimulation-to-torque models only reflect that portion of the neural excitation that results directly from stimulation. The time-varying properties of the excitation process cannot be captured by existing stimulation-to-torque models, but they are tracked by the EEMG-to-torque models that are developed here. This work represents a promising approach to the real-time estimation of stimulated muscle force in functional neuromuscular stimulation applications.

Closed-loop drug infusion for control of heart-rate trajectory in pharmacological stress tests.

The diagnosis of coronary artery disease (CAD) is an important task in the management of cardiology patients. Recently, the use of pharmacological stress testing has become available as an alternative to exercise stress testing (ETT). A new system (device-drug combination) was developed specifically for the diagnosis of coronary artery disease. The system uses a novel catecholamine, arbutamine, which is infused intravenously to increase heart rate (HR) and cardiac contractility in order to evoke signs of ischemia. The development of a closed-loop control algorithm for the delivery of this drug and a pharmacodynamic (PD) model representing the HR response to arbutamine infusions are presented. Model parameters are estimated from clinical data on normal volunteers and patients. Based on this mathematical model, a rule-based control algorithm is designed. The structure of the control algorithm is discussed and testing of the algorithm based on simulations and animal and human trials are summarized. Results from clinical trials shows that the algorithm controls the HR increase according to a selected trajectory. The automated delivery of the drug can provide the cardiologist with an efficient, effective, and safe method for administering a pharmacological stress test.

Estimating mechanical parameters of leg segments in individuals with and without physical disabilities.

Methods are described for estimating the inertia, viscosity, and stiffness of the lower leg around the knee and of the whole leg around the hip that are applicable even to persons with considerable spasticity. These involve: 1) a "pull" test in which the limb is slowly moved throughout its range of motion while measuring angles (with an electrogoniometer) and torques (with a hand-held dynamometer) to determine passive stiffness and 2) a "pendulum" test in which the limb is moved against gravity and then dropped, while again measuring angles and torques. By limiting the extent of the movement and choosing a direction (flexion or extension) that minimizes reflex responses, the mechanical parameters can be determined accurately and efficiently using computer programs. In the sample of subjects studied (nine with disability related to spinal cord injury, head injury, or stroke, and nine with no neurological disability), the inertia of the lower leg was significantly reduced in the subjects with disability (p < 0.05) as a result of atrophy, but the stiffness and viscosity were within normal limits. The values of inertia were also compared with anthropometric data in the literature. The identification of these passive parameters is particularly important in designing systems for functional electrical stimulation of paralyzed muscles, but the methods may be widely applicable in rehabilitation medicine.

On the existence of a lactate threshold during incremental exercise: a systems analysis.

The relationship between blood lactate concentration ([La]) and O2 uptake (VO2) during incremental exercise remains controversial: does [La] increase smoothly as a function of VO2 (continuous model), or does it begin to increase abruptly above a particular metabolic rate (threshold model)? The dynamic characteristics of the underlying physiological system are investigated using system identification analysis techniques. A multivariate deterministic time series model of the [La] and VO2 response to incremental changes in work rate was fitted to simulated and experimental data. Time-varying system response parameters were determined through the application of a weighted recursive least squares algorithm. The model, using the identified time-varying parameters, provided a good fit to the data. The variation of these parameters over time was then examined. Two major transitions in the parameters were found to occur at intensity levels equivalent to 53 +/- 8% and 77 +/- 9% maximal VO2 (experimental data). These changes in the model parameters indicate that the best linear dynamic model that fits the observed system behavior has changed. This implies that the system has changed its operation in some way, by altering its structure or by moving to a different operating region. The identified parameter changes over time suggest that the exercise intensity range (from rest to maximal VO2) is divided into three main intensity domains, each with distinct dynamics. Further study of this three-phase system may help in the understanding of the underlying physiological mechanisms that affect the dynamics of [La] and VO2 during exercise.

Neural network control of functional neuromuscular stimulation systems: computer simulation studies.

A neural network control system has been designed for the control of cyclic movements in Functional Neuromuscular Stimulation (FNS) systems. The design directly addresses three major problems in FNS control systems: customization of control system parameters for a particular individual, adaptation during operation to account for changes in the musculoskeletal system, and attaining resistance to mechanical disturbances. The control system was implemented by a two-stage neural network that utilizes a combination of adaptive feedforward and feedback control techniques. A new learning algorithm was developed to provide rapid customization and adaptation. The control system was evaluated in a series of studies on a computer simulated musculoskeletal model. The model of electrically stimulated muscle used in the study included nonlinear recruitment, linear dynamics, and multiplicative nonlinear torque-angle and torque-velocity scaling factors. The skeletal model consisted of a one-segment planar system with passive constraints on joint movement. Results of the evaluation have demonstrated that the control system can provide automated customization of the feedforward controller parameters for a given musculoskeletal system. It can account for changes in the musculoskeletal system by adapting the feedforward controller parameters on-line and it can resist the effects of mechanical disturbances. These results suggest that this design may be suitable for the control of FNS systems and other dynamic systems.

Muscle-joint models incorporating activation dynamics, moment-angle, and moment-velocity properties.

Muscle input/output models incorporating activation dynamics, moment-angle, and moment-velocity factors are commonly used to predict the moment produced by muscle during nonisometric contractions; the three factors are generally assumed to be independent. We examined the ability of models with independent factors, as well as models with coupled factors, to fit input/output data measured during simultaneous modulation of the fraction of muscle stimulated (recruitment) and joint angle inputs. The models were evaluated in stimulated cat soleus muscles producing ankle extension moment, with regard to their potential applications in neuroprostheses with either fixed parameters or parameter adaptation. Both uncoupled and coupled models predicted the output moment well for random angle perturbation sizes ranging from 10 degrees to 30 degrees. For the uncoupled model, the best parameter values depended on the range of perturbations and the mean angle. Introducing coupling between activation and velocity in the model reduced this parameter sensitivity; one set of model parameter values fit the data for all perturbation sizes and also fit the data under isometric or constant stimulation conditions. Thus, the coupled model would be the most appropriate for applications requiring fixed parameter values. In contrast, with continuous parameter adaptation, errors due to changing test conditions decreased more quickly for the uncoupled model, suggesting that it would perform well in adaptive control of neuroprostheses.

Adaptive buffering of breath-by-breath variations of end-tidal CO2 of humans.

We have designed and implemented a computer-controlled system that uses an adaptive control algorithm (generalized minimum variance) to buffer the breath-by-breath variations of the end-tidal CO2 fraction (FETCO2) that occur spontaneously or are exaggerated in certain experimental protocols (e.g., induced hypoxia, any type of induced variations in the ventilatory pattern). Near the end of each breath, FETCO2 of the following breath is predicted and the inspired CO2 fraction (FICO2) of the upcoming breath is adjusted to minimize the difference between the predicted and desired FETCO2 of the next breath. The one-breath-ahead prediction of FETCO2 is based on an adaptive autoregressive with exogenous inputs (ARX) model: FETCO2 of a given breath is related to FICO2, FETCO2 of the previous breath, and inspiratory ventilation. Adequacy of the prediction is demonstrated using data from experiments in which FICO2 was varied pseudorandomly in wakefulness and sleep. The algorithm for optimally buffering changes in FETCO2 is based on the coefficients of the ARX model. We have determined experimentally the frequency of FETCO2 variations that can be buffered adequately by our controller, testing both spontaneous variations in FETCO2 and variations induced by hypoxia in young awake human subjects. The controller is most effective in buffering variations of FETCO2 in the frequency range of <0.1 cycle/breath. Some potential applications are discussed.

Nonlinear joint angle control for artificially stimulated muscle.

Designs of both open- and closed-loop controllers of electrically stimulated muscle that explicitly depend on a nonlinear mathematical model of muscle input-output properties are presented and evaluated. The muscle model consists of three factors: a muscle activation dynamics factor, an angle-torque relationship factor, and an angular velocity torque relationship factor. These factors are multiplied to relate output torque to input stimulation and joint angle. An experimental method for the determination of the parameters of this model was designed, implemented, and evaluated. An open-loop nonlinear compensator, based upon this model, was tested in an animal model. Its performance in the control of joint angle in the presence of a known load was compared with a PID controller, and with a combination of the PID controller and the nonlinear compensator. The performance of the nonlinear compensator appeared to be strongly dependent on modeling errors. Its performance was roughly equivalent to that of the PID controller alone: somewhat better when the model was accurate, and somewhat worse when it was inaccurate. Combining the nonlinear open loop compensator with the PID feedback controller improved performance when the model was accurate.

Instrumented parallel bars for three-dimensional force measurement.

This paper describes the modification and instrumentation of standard parallel bars to allow for the measurement of applied forces on both horizontal bars in three dimensions. This measurement system has been used in the development and evaluation of functional electrical stimulation (FES) devices for standing and gait restoration in paralyzed patients. Real-time measurement of forces applied by the upper body of the patient to the parallel bars is of use in the evaluation of FES stimulation patterns (or automatic controllers of stimulation). Such measurements are useful in the redesign of stimulation patterns and/or stimulation controllers.

Feedback control methods for task regulation by electrical stimulation of muscles.

Three feedback control algorithms of varying complexity were compared for controlling three different tasks during electrical stimulation of muscles. Two controllers use stimulus pulse width (or recruitment) modulation to grade muscle force (the fixed parameter, first-order PW controller and the adaptive controller). The third controller varies both stimulus pulse width and period simultaneously for muscle force modulation (the PW/SP controller described in the comparison paper). The three tasks tested were isometric torque control, unloaded position tracking, and control of transitions between isometric and unloaded conditions. The first task involved the muscle recruitment nonlinearity. The second task added the effects of muscle length-tension and force-velocity nonlinearities. The third task included a sudden changes in external loading conditions. The comparative evaluation was carried out in an intact cat ankle joint with stimulation of tibialis anterior and medial gastrocnemius muscles. The simplest PW controller demonstrated robust control for all tasks. The PW/SP controller improved the performance of the PW controller significantly for control of isometric torque and load transition, but only slightly for control of unloaded joint position. However, the adaptive controller did not consistently achieve a significant improvement in performance compared with the PW controller for any task. Results suggest that muscle length-tension and force-velocity nonlinearities affect the performance of these controllers similarly within the tested ranges of movement amplitudes and speeds. Abrupt changes in the system, such as those due to recruitment nonlinearity and external loading transitions, tend to limit the performance of the adaptive controller. The study provides guidelines for choosing control algorithms for neural prostheses.

Feedback control of electrically stimulated muscle using simultaneous pulse width and stimulus period modulation.

This paper considers the closed-loop control of electrically stimulated muscle using simultaneous pulse width and frequency modulation. Previous work has experimentally demonstrated good feedback regulation of muscle force using fixed parameter and an adaptive controller modulating pulse width. In this work, it is shown how the addition of pulse frequency modulation to pulse width modulation can improve controller performance. This combination controller has been developed for both single muscle activation and for costimulation of antagonists. This is accomplished using a single command input. In single muscle operation, the combination of pulse width and stimulus pulse frequency modulation results in better control of transient responses than with pulse width modulation alone; the total number of stimulus pulses is increased, however, when compared with pulse width-only modulation at the muscle fusion frequency. In the case of costimulation, the controller modulates the pulse stimulus periods of the antagonists in a reciprocal manner, to ensure stable and fast responses. That is, the frequency of stimulation of the antagonist is increased when that of the agonist is decreased. This results in better control performance with generally fewer stimulus pulses than those generated by costimulation using only pulse width modulation. This feedback controller was evaluated in animal experiments. Step responses with rapid rise times but without overshoot were obtained by the combined modulation. Good steady-state and transient performance were obtained over a wide range of static lengths and commands, under different loading conditions and in different animals. This controller is a promising potential component of neural prostheses to restore functional movement in paralyzed individuals.

Control of end-point forces of a multijoint limb by functional neuromuscular stimulation.

A multivariable feedback controller was designed and tested for regulating the magnitude and orientation of the force vector at the end point of a multijoint limb in contact with an isometric load. The force vector was produced by electrical stimulation of muscles. To achieve arbitrary control of end-point force magnitude and orientation, two coupling issues must be dealt with by the control system. First, there is a geometric coupling between the end-point force vector and joint torques. The amplitude and orientation of the force vector depend on the limb geometry. Second, torques at two joints may be coupled due to activation of muscles that cross them (biarticular coupling). To eliminate the geometric coupling, a transformation of controller error from the Cartesian space to the joint space was employed. A multivariable proportional-plus-integral (PI) control law was used to calculate muscle activation based on the transformed controller error. Centralized and decentralized controls were investigated for decoupling the effects of biarticular muscles. The results obtained from cat experiments showed that the magnitude and orientation of the end-point forces of the cat hindlimb could be regulated by this controller. In the presence of strong biarticular coupling, centralized control yielded better performance than decentralized control during transient responses. Both control strategies could decouple the biarticular muscle at steady state. When no biarticular coupling was present, centralized control sometimes performed worse than decentralized control. This is the first step in the simultaneous control of multiple joints by functional neuromuscular stimulation (FNS). The controller has broad potential applications in FNS neural prostheses.

Feedback control of coronal plane hip angle in paraplegic subjects using functional neuromuscular stimulation.

This paper reports on an investigation of feedback control of coronal plane posture in paraplegic subjects who stand using functional neuromuscular stimulation (FNS). A feedback control system directed at regulating coronal plane hip angle in neutral position was designed, implemented, and evaluated in two paraplegic subjects. The control system included sensor mounting and signal processing techniques, a two-stage feedback controller, stimulation hardware, and a set of percutaneous intramuscular electrodes. The feedback controller consisted of two-stages in cascade: a modified discrete-time proportional-integral-derivative (PID) stage and a nonlinear single-input, multiple-output stage to determine the stimulation to be sent to several muscles. The focus of this work was on evaluating the performance of the feedback controller by comparing the response of the feedback-controlled system to that of an open-loop stimulation system. In an evaluation based on temporal response characteristics the controlled system exhibited a 41% reduction in root-mean-squared (rms) error (where error is defined as the deviation from the desired angle), a 52% reduction in steady-state error, and a 22% reduction in hip compliance. In addition, the feedback-controlled system exhibited significant reductions in variability of these measures on several days. These results demonstrate the ability of the feedback controller to improve the temporal response characteristics of the FNS control system.

Recursive parameter identification of constrained systems: an application to electrically stimulated muscle.

In the application of real-time identification methods for diagnosis or adaptive control of biomedical systems, there is often known model information that is ignored. Constraints on the allowable values of parameters, which may be based on physical considerations, are often neglected because the information does "fit" easily into commonly used parameter-identification algorithms. In this paper a method of incorporating constraints on model parameters is developed. This method is applicable to most recursive parameter-identification algorithms. It enforces linear equality constraints on identified parameters. The use of this method for the real-time identification of autoregressive moving-average-type time series models, subject to parameter constraints, is described in detail. These constraints may be time varying. At each time step, the parameter estimate obtained by a recursive least squares estimator is orthogonally projected onto the constraint surface. This simple idea, when appropriately executed, enhances the output prediction accuracy of estimated parameters. Using constraint information in this way is important when we do not wish to destroy a "natural" parameterization of the model (by an initial projection to incorporate equality constraints), or when we cannot use a single initial model simplification (because the constraints are time varying or involve inputs and outputs). Because it improves output prediction at future times, this method is advantageous for use in predictive adaptive controllers. The use of this algorithm is demonstrated in the identification of electrically stimulated quadriceps muscles in paraplegic human subjects, using percutaneous intramuscular electrodes. The nonlinear steady-state force versus pulsewidth recruitment characteristic of the electrode-muscle system is identified simultaneously with the input-output muscle response dynamics, using a Hammerstein-type model. Knowledge of the recruitment curve's shape is translated into constraints on the identified parameters. This information improves the experimental predictive quality of the identified model.

Adaptive control is enhanced by background estimation.

The automated control of physiological variables must often contend with an unknown and time-varying background (i.e., the output level corresponding to no input). To allow for simultaneous real-time identification of background as well as the parameters of an autoregressive moving average model with exogenous inputs (ARMAX model) during adaptive control, a "floating identifier" (FI) approach was developed which may be used with most recursive identification algorithms. This method separates input and output data into low- and high-frequency components. The high-frequency components are used to identify the ARMAX model parameters and the low-frequency components to identify background. This approach was evaluated in computer simulations and animal experiments comparing an adaptive controller coupled to the FI with the same controller coupled to two other standard least squares identifiers. In the animal experiments, sodium nitroprusside was used to control mean arterial pressure of anesthetized dogs in the presence of background changes. Results showed that with the FI, the controller performed satisfactorily, while with the other identifiers, it sometimes failed. It is concluded that the FI approach is useful when applying ARMAX-based adaptive controllers to systems in which a change in background is likely.

Feedback regulation of hand grasp opening and contact force during stimulation of paralyzed muscle.

A fixed-parameter, discrete-time, first-order, feedback control system is described for regulating grasp during electrical stimulation of paralyzed muscles of the hand. The stiffness of the grasp (relationship between grasp force and grasp opening) is kept constant by linearly combining force and position feedback signals. Thus, a single continuous command signal can control the size of the grasp opening prior to object acquisition and both grasp force and opening after contact. The controller achieves this change in controlled variables by scaling and summing the force and position feedback signals, rather than by a discrete switch in control strategy. Experimental tests of the control system in quadriplegic subjects show that control can be obtained over conditions ranging from unloaded position regulation to isometric force regulation, as well as in the transition between these conditions. The robustness of the control system was evaluated during force regulation with isometric loads. Step response rise time and overshoot were much more dependent on system gain than on the location of the controller zero. Responses with rise time less than two seconds and overshoot less than 30% were obtained over a gain range up to ten, indicating good robustness to muscle gain reductions such as might be caused by fatigue.

Tetanic responses of electrically stimulated paralyzed muscle at varying interpulse intervals.

The influence of stimulus interpulse interval (IPI) on torque output during electrically-evoked contractions was investigated for the knee extensor muscles of paralyzed subjects. The parameters measured were the rise time, magnitude, and relaxation time of the contraction at stimulus IPI's ranging from 62 to 7 ms. Torque output increased as IPI's were decreased from 62 to 15 ms. Peak torques were recorded at IPI's of 12-15 ms; IPI's less than these resulted in an insignificant loss of torque. Rise times decreased as IPI's were decreased. Relaxation time generally increased as IPI's were decreased with the longest relaxation times occurring with stimulation at an IPI of 12 ms. To demonstrate the influence of IPI on muscle fatigue, the effect of prolonged stimulation at short (12 ms) and long (50 ms) IPI's was also compared. After 30 s of stimulation with an IPI of 12 ms, mean torque had declined to 5 +/- 3 percent and after 30 s of stimulation with an IPI of 50 ms, mean torque had declined to 82 +/- 4 percent of the initial value. Knowledge of how stimulus IPI influences the response of paralyzed muscle to electrical stimulation may assist in the development of rehabilitation devices which utilize these technologies.