Rehabilitation Assessment System for Stroke Patients Based on Fusion-Type Optoelectronic Plethysmography Device and Multi-Modality Fusion Model: Design and Validation.

This study aimed to propose a portable and intelligent rehabilitation evaluation system for digital stroke-patient rehabilitation assessment. Specifically, the study designed and developed a fusion device capable of emitting red, green, and infrared lights simultaneously for photoplethysmography (PPG) acquisition. Leveraging the different penetration depths and tissue reflection characteristics of these light wavelengths, the device can provide richer and more comprehensive physiological information. Furthermore, a Multi-Channel Convolutional Neural Network-Long Short-Term Memory-Attention (MCNN-LSTM-Attention) evaluation model was developed. This model, constructed based on multiple convolutional channels, facilitates the feature extraction and fusion of collected multi-modality data. Additionally, it incorporated an attention mechanism module capable of dynamically adjusting the importance weights of input information, thereby enhancing the accuracy of rehabilitation assessment. To validate the effectiveness of the proposed system, sixteen volunteers were recruited for clinical data collection and validation, comprising eight stroke patients and eight healthy subjects. Experimental results demonstrated the system's promising performance metrics (accuracy: 0.9125, precision: 0.8980, recall: 0.8970, F1 score: 0.8949, and loss function: 0.1261). This rehabilitation evaluation system holds the potential for stroke diagnosis and identification, laying a solid foundation for wearable-based stroke risk assessment and stroke rehabilitation assistance.

Continuous Motion Intention Prediction Using sEMG for Upper-Limb Rehabilitation: A Systematic Review of Model-Based and Model-Free Approaches.

Upper limb functional impairments persisting after stroke significantly affect patients' quality of life. Precise adjustment of robotic assistance levels based on patients' motion intentions using sEMG signals is crucial for active rehabilitation. This paper systematically reviews studies on continuous prediction of upper limb single joints and multi-joint combinations motion intention using Model-Based (MB) and Model-Free (MF) approaches over the past decade, based on 186 relevant studies screened from six major electronic databases. The findings indicate ongoing challenges in terms of subject composition, algorithm robustness and generalization, and algorithm feasibility for practical applications. Moreover, it suggests integrating the strengths of both MB and MF approaches to improve existing algorithms. Therefore, future research should further explore personalized MB-MF combination methods incorporating deep learning, attention mechanisms, muscle synergy features, motor unit features, and closed-loop feedback to achieve precise, real-time, and long-duration prediction of multi-joint complex movements, while further refining the transfer learning strategy for rapid algorithm deployment across days and subjects. Overall, this review summarizes the current research status, significant findings, and challenges, aiming to inspire future research on predicting upper limb motion intentions based on sEMG.

Simultaneous Hip Implant Segmentation and Gruen Landmarks Detection.

The assessment of implant status and complications of Total Hip Replacement (THR) relies mainly on the clinical evaluation of the X-ray images to analyse the implant and the surrounding rigid structures. Current clinical practise depends on the manual identification of important landmarks to define the implant boundary and to analyse many features in arthroplasty X-ray images, which is time-consuming and could be prone to human error. Semantic segmentation based on the Convolutional Neural Network (CNN) has demonstrated successful results in many medical segmentation tasks. However, these networks cannot define explicit properties that lead to inaccurate segmentation, especially with the limited size of image datasets. Our work integrates clinical knowledge with CNN to segment the implant and detect important features simultaneously. This is instrumental in the diagnosis of complications of arthroplasty, particularly for loose implant and implant-closed bone fractures, where the location of the fracture in relation to the implant must be accurately determined. In this work, we define the points of interest using Gruen zones that represent the interface of the implant with the surrounding bone to build a Statistical Shape Model (SSM). We propose a multitask CNN that combines regression of pose and shape parameters constructed from the SSM and semantic segmentation of the implant. This integrated approach has improved the estimation of implant shape, from 74% to 80% dice score, making segmentation realistic and allowing automatic detection of Gruen zones. To train and evaluate our method, we generated a dataset of annotated hip arthroplasty X-ray images that will be made available.

Detecting latent topics and trends of digital twins in healthcare: A structural topic model-based systematic review.

Objective: Digital twins (DTs) have received widespread attention recently, providing new ideas and possibilities for future healthcare. This review aims to provide a quantitative review to analyze specific study contents, research focus, and trends of DT in healthcare. Simultaneously, this review intends to expand the connotation of "healthcare" into two directions, namely "Disease treatment" and "Health enhancement" to analyze the content within the "DT + healthcare" field thoroughly. Methods: A data mining method named Structure Topic Modeling (STM) was used as the analytical tool due to its topic analysis ability and versatility. Google Scholar, Web of Science, and China National Knowledge Infrastructure supplied the material papers in this review. Results: A total of 94 high-quality papers published between 2018 and 2022 were gathered and categorized into eight topics, collectively covering the transformative impact across a broader spectrum in healthcare. Three main findings have emerged: (1) papers published in healthcare predominantly concentrate on technology development (artificial intelligence, Internet of Things, etc.) and application scenarios(personalized, precise, and real-time health service); (2) the popularity of research topics is influenced by various factors, including policies, COVID-19, and emerging technologies; and (3) the preference for topics is diverse, with a general inclination toward the attribute of "Health enhancement." Conclusions: This review underscores the significance of real-time capability and accuracy in shaping the future of DT, where algorithms and data transmission methods assume central importance in achieving these goals. Moreover, technological advancements, such as omics and Metaverse, have opened up new possibilities for DT in healthcare. These findings contribute to the existing literature by offering quantitative insights and valuable guidance to keep researchers ahead of the curve.

SSVEP-Based Brain-Computer Interface Controlled Robotic Platform With Velocity Modulation.

Steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) have been extensively studied due to many benefits, such as non-invasiveness, high information transfer rate, and ease of use. SSVEP-based BCI has been investigated in various applications by projecting brain signals to robot control commands. However, the movement direction and speed are generally fixed and prescribed, neglecting the user's requirement for velocity changes during practical implementations. In this study, we proposed a velocity modulation method based on stimulus brightness for controlling the robotic arm in the SSVEP-based BCI system. A stimulation interface was designed, incorporating flickers, target and a cursor workspace. The synchronization of the cursor and robotic arm does not require the subject's eye switch between the stimuli and the robot. The feature vector consists of the characteristics of the signal and the classification result. Subsequently, the Gaussian mixture model (GMM) and Bayesian inference were used to calculate the posterior probabilities that the signal came from a high or low brightness flicker. A brain-actuated speed function was designed by incorporating the posterior probability difference. Finally, the historical velocity was considered to determine the final velocity. To demonstrate the effectiveness of the proposed method, online experiments, including single- and multi-target reaching tasks, were conducted. The extensive experimental results validated the feasibility of the proposed method in reducing reaching time and achieving proximity to the target.

LSTM-AE for Domain Shift Quantification in Cross-Day Upper-Limb Motion Estimation Using Surface Electromyography.

Although deep learning (DL) techniques have been extensively researched in upper-limb myoelectric control, system robustness in cross-day applications is still very limited. This is largely caused by non-stable and time-varying properties of surface electromyography (sEMG) signals, resulting in domain shift impacts on DL models. To this end, a reconstruction-based method is proposed for domain shift quantification. Herein, a prevalent hybrid framework that combines a convolutional neural network (CNN) and a long short-term memory network (LSTM), i.e. CNN-LSTM, is selected as the backbone. The paring of auto-encoder (AE) and LSTM, abbreviated as LSTM-AE, is proposed to reconstruct CNN features. Based on reconstruction errors (RErrors) of LSTM-AE, domain shift impacts on CNN-LSTM can be quantified. For a thorough investigation, experiments were conducted in both hand gesture classification and wrist kinematics regression, where sEMG data were both collected in multi-days. Experiment results illustrate that, when the estimation accuracy degrades substantially in between-day testing sets, RErrors increase accordingly and can be distinct from those obtained in within-day datasets. According to data analysis, CNN-LSTM classification/regression outcomes are strongly associated with LSTM-AE errors. The average Pearson correlation coefficients could reach -0.986 ± 0.014 and -0.992 ± 0.011, respectively.

Cross-Subject Transfer Learning for Boosting Recognition Performance in SSVEP-based BCIs.

Steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) have been substantially studied in recent years due to their fast communication rate and high signal-to-noise ratio. The transfer learning is typically utilized to improve the performance of SSVEP-based BCIs with auxiliary data from the source domain. This study proposed an inter-subject transfer learning method for enhancing SSVEP recognition performance through transferred templates and transferred spatial filters. In our method, the spatial filter was trained via multiple covariance maximization to extract SSVEP-related information. The relationships between the training trial, the individual template, and the artificially constructed reference are involved in the training process. The spatial filters are applied to the above templates to form two new transferred templates, and the transferred spatial filters are obtained accordingly via the least-square regression. The contribution scores of different source subjects can be calculated based on the distance between the source subject and the target subject. Finally, a four-dimensional feature vector is constructed for SSVEP detection. To demonstrate the effectiveness of the proposed method, a publicly available dataset and a self-collected dataset were employed for performance evaluation. The extensive experimental results validated the feasibility of the proposed method for improving SSVEP detection.

Actuation system modelling and design optimization for an assistive exoskeleton for disabled and elderly with series and parallel elasticity.

BACKGROUND: The aim of a robotic exoskeleton is to match the torque and angular profile of a healthy human subject in performing activities of daily living. Power and mass are the main requirements considered in the robotic exoskeletons that need to be reduced so that portable designs to perform independent activities by the elderly users could be adopted. OBJECTIVE: This paper evaluates a systematic approach for the design optimization strategies of elastic elements and implements an actuator design solution for an ideal combination of components of an elastic actuation system while providing the same level of support to the elderly. METHODS: A multi-factor optimization technique was used to determine the optimum stiffness and engagement angle of the spring within its elastic limits at the hip, knee and ankle joints. An actuator design framework was developed for the elderly users to match the torque-angle characteristics of the healthy human with the best motor and transmission system combined with series or parallel elasticity in an elastic actuator. RESULTS: With the optimized spring stiffness, a parallel elastic element significantly reduced the torque and power requirements up to 90% for some manoeuvres for the users to perform ADL. When compared with the rigid actuation system, the optimized robotic exoskeleton actuation system reduced the power consumption of up to 52% using elastic elements. CONCLUSION: A lightweight, smaller design of an elastic actuation system consuming less power as compared to a rigid system was realized using this approach. This will help to reduce the battery size and hence the portability of the system could be better adopted to support elderly users in performing daily living activities. It was established that parallel elastic actuators (PEA) can reduce the torque and power better than series elastic actuators (SEA) in performing everyday tasks for the elderly.

TEleRehabilitation Nepal (TERN) for People With Spinal Cord Injury and Acquired Brain Injury: A Feasibility Study.

BACKGROUND: Spinal Cord Injury (SCI) or Acquired Brain Injury (ABI) leads to disability, unemployment, loss of income, decreased quality of life and increased mortality. The impact is worse in Low-and Middle-Income Countries (LMICs) due to a lack of efficient long-term rehabilitative care. This study aims to explore the feasibility and acceptability of a telerehabilitation programme in Nepal. METHODS: Prospective cohort feasibility study in a community setting following discharge from a specialist rehabilitation centre in Nepal. Patients with SCI or ABI who had previously accessed specialist rehabilitation were connected to a specialist Multidisciplinary Team (MDT) in the centre through a video conference system for comprehensive remote assessments and virtual individualised interventions. Data were captured on recruitment, non-participation rates, retention, acceptability (via end-of-study in-depth interviews with a subset of participants) and outcome measures including the Modified Barthel Index (MBI), Depression Anxiety Stress Scale (DASS) and EuroQol-5D (EQ-5D), completed pre- and post-programme. RESULTS: 97 participants with SCI (n = 82) or ABI (n = 15) discharged from the centre during an 18-month period were approached and enrolled on the study. The telerehabilitation programme facilitated the delivery of support around multiple aspects of rehabilitation care, such as spasticity treatments and pain management. Outcome measures indicated a significant improvement in functional independence (P < .001), depression, anxiety and stress (P < .001) and quality of life (P < .001). Qualitative interviews (n = 18) revealed participants found the programme acceptable, valuing regular contact and input from MDT professionals and avoiding expensive and lengthy travel. CONCLUSION: This is the first study in Nepal to identify telerehabilitation as a feasible and acceptable approach to augment the provision of specialist rehabilitation. Future research is needed to assess the suitability of the programme for other conditions requiring specialist rehabilitation and determine the mechanisms underpinning improved outcomes for people with SCI or ABI. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT04914650.

Toward Robust, Adaptiveand Reliable Upper-Limb Motion Estimation Using Machine Learning and Deep Learning-A Survey in Myoelectric Control.

To develop multi-functionalhuman-machine interfaces that can help disabled people reconstruct lost functions of upper-limbs, machine learning (ML) and deep learning (DL) techniques have been widely implemented to decode human movement intentions from surface electromyography (sEMG) signals. However, due to the high complexity of upper-limb movements and the inherent non-stable characteristics of sEMG, the usability of ML/DL based control schemes is still greatly limited in practical scenarios. To this end, tremendous efforts have been made to improve model robustness, adaptation, and reliability. In this article, we provide a systematic review on recent achievements, mainly from three categories: multi-modal sensing fusion to gain additional information of the user, transfer learning (TL) methods to eliminate domain shift impacts on estimation models, and post-processing approaches to obtain more reliable outcomes. Special attention is given to fusion strategies, deep TL frameworks, and confidence estimation. Research challenges and emerging opportunities, with respect to hardware development, public resources, and decoding strategies, are also analysed to provide perspectives for future developments.

Automatic detection and classification of peri-prosthetic femur fracture.

PURPOSE: Object classification and localization is a key task of computer-aided diagnosis (CAD) tool. Although there have been numerous generic deep learning (DL) models developed for CAD, there is no work in the literature to evaluate their effectiveness when utilized in diagnosing fractures in proximity of joint implants. In this work, we aim to assess the performance of existing classification systems on binary and multi-class problems (fracture types) using plain radiographs. In addition, we evaluated the performance of object detection systems using the one- and two-stage DL architectures. METHODS: A data set of 1272 X-ray images of Peri-prosthetic Femur Fracture PFF was collected. The fractures were annotated with bounding boxes and classified according to the Vancouver Classification System (type A, B, C) by two clinical specialists. Four classification models such as Densenet161, Resnet50, Inception, VGG and two object detection models such as Faster RCNN and RetinaNet were evaluated, and their performance compared. Six confusion matrix-based measures were reported to evaluate fracture classification. For localization of the fracture, Average Precision and localization accuracy were reported. RESULTS: The Resnet50 showed the best performance with [Formula: see text] accuracy and [Formula: see text] F1-score in the binary classification: fracture/normal. In addition, the Resnet50 showed [Formula: see text] accuracy in multi-classification (normal, Vancouver type A, B and C). CONCLUSIONS: A large data set of PFF images and the annotations of fracture features by two independent assessments were created to implement a DL-based approach for detecting, classifying and localizing PFFs. It was shown that this approach could be a promising diagnostic tool of fractures in proximity of joint implants.

Physics-informed Deep Learning for Musculoskeletal Modelling: Predicting Muscle Forces and Joint Kinematics from Surface EMG.

Musculoskeletal models have been widely used for detailed biomechanical analysis to characterise various functional impairments given their ability to estimate movement variables (i.e., muscle forces and joint moments) which cannot be readily measured in vivo. Physics-based computational neuromusculoskeletal models can interpret the dynamic interaction between neural drive to muscles, muscle dynamics, body and joint kinematics and kinetics. Still, such set of solutions suffers from slowness, especially for the complex models, hindering the utility in real-time applications. In recent years, data-driven methods have emerged as a promising alternative due to the benefits in speedy and simple implementation, but they cannot reflect the underlying neuromechanical processes. This paper proposes a physics-informed deep learning framework for musculoskeletal modelling, where physics-based domain knowledge is brought into the data-driven model as soft constraints to penalise/regularise the data-driven model. We use the synchronous muscle forces and joint kinematics prediction from surface electromyogram (sEMG) as the exemplar to illustrate the proposed framework. Convolutional neural network (CNN) is employed as the deep neural network to implement the proposed framework. Simultaneously, the physics law between muscle forces and joint kinematics is used the soft constraint. Experimental validations on two groups of data, including one benchmark dataset and one self-collected dataset from six healthy subjects, are performed. The experimental results demonstrate the effectiveness and robustness of the proposed framework.

A review of gait disorders in the elderly and neurological patients for robot-assisted training.

Purpose: Ambulation is an important objective for people with pathological gaits. Exoskeleton robots can assist these people to complete their activities of daily living. There are exoskeletons that have been presented in literature to assist the elderly and other pathological gait users. This article presents a review of the degree of support required in the elderly and neurological gait disorders found in the human population. This will help to advance the design of robot-assisted devices based on the needs of the end users.Methods: The articles included in this review are collected from different databases including Science Direct, Springer Link, Web of Science, Medline and PubMed and with the purpose to investigate the gait parameters of elderly and neurological patients. Studies were included after considering the full texts and only those which focus on spatiotemporal, kinematic and kinetic gait parameters were selected as they are most relevant to the scope of this review. A systematic review and meta-analysis were conducted.Results: The meta-analysis report on the spatiotemporal, kinematic and kinetic gait parameters of elderly and neurological patients revealed a significant difference based on the type and level of impairment. Healthy elderly population showed deviations in the gait parameters due to age, however, significant difference is observed in the gait parameters of the neurological patients.Conclusion: A level of agreement was observed in most of the studies however the review also noticed some controversies among different studies in the same group. The review on the spatiotemporal, kinematics and kinetic gait parameters will provide a summary of the fundamental needs of the users for the future design and development of robotic assistive devices.Implications for rehabilitationThe support requirements provide the foundation for designing assistive devices.The findings will be crucial in defining the design criteria for robot assistive devices.

A new multivariate empirical mode decomposition method for improving the performance of SSVEP-based brain-computer interface.

OBJECTIVE: Accurate and efficient detection of steady-state visual evoked potentials (SSVEP) in electroencephalogram (EEG) is essential for the related brain-computer interface (BCI) applications. APPROACH: Although the canonical correlation analysis (CCA) has been applied extensively and successfully to SSVEP recognition, the spontaneous EEG activities and artifacts that often occur during data recording can deteriorate the recognition performance. Therefore, it is meaningful to extract a few frequency sub-bands of interest to avoid or reduce the influence of unrelated brain activity and artifacts. This paper presents an improved method to detect the frequency component associated with SSVEP using multivariate empirical mode decomposition (MEMD) and CCA (MEMD-CCA). EEG signals from nine healthy volunteers were recorded to evaluate the performance of the proposed method for SSVEP recognition. MAIN RESULTS: We compared our method with CCA and temporally local multivariate synchronization index (TMSI). The results suggest that the MEMD-CCA achieved significantly higher accuracy in contrast to standard CCA and TMSI. It gave the improvements of 1.34%, 3.11%, 3.33%, 10.45%, 15.78%, 18.45%, 15.00% and 14.22% on average over CCA at time windows from 0.5 s to 5 s and 0.55%, 1.56%, 7.78%, 14.67%, 13.67%, 7.33% and 7.78% over TMSI from 0.75 s to 5 s. The method outperformed the filter-based decomposition (FB), empirical mode decomposition (EMD) and wavelet decomposition (WT) based CCA for SSVEP recognition. SIGNIFICANCE: The results demonstrate the ability of our proposed MEMD-CCA to improve the performance of SSVEP-based BCI.

Musculoskeletal modelling of human ankle complex: Estimation of ankle joint moments.

BACKGROUND: A musculoskeletal model for the ankle complex is vital in order to enhance the understanding of neuro-mechanical control of ankle motions, diagnose ankle disorders and assess subsequent treatments. Motions at the human ankle and foot, however, are complex due to simultaneous movements at the two joints namely, the ankle joint and the subtalar joint. The musculoskeletal elements at the ankle complex, such as ligaments, muscles and tendons, have intricate arrangements and exhibit transient and nonlinear behaviour. METHODS: This paper develops a musculoskeletal model of the ankle complex considering the biaxial ankle structure. The model provides estimates of overall mechanical characteristics (motion and moments) of ankle complex through consideration of forces applied along ligaments and muscle-tendon units. The dynamics of the ankle complex and its surrounding ligaments and muscle-tendon units is modelled and formulated into a state space model to facilitate simulations. A graphical user interface is also developed during this research in order to include the visual anatomical information by converting it to quantitative information on coordinates. FINDINGS: Validation of the ankle model was carried out by comparing its outputs with those published in literature as well as with experimental data obtained from an existing parallel ankle rehabilitation robot. INTERPRETATION: Qualitative agreement was observed between the model and measured data for both, the passive and active ankle motions during trials in terms of displacements and moments.

A MUSIC-based method for SSVEP signal processing.

The research on brain computer interfaces (BCIs) has become a hotspot in recent years because it offers benefit to disabled people to communicate with the outside world. Steady state visual evoked potential (SSVEP)-based BCIs are more widely used because of higher signal to noise ratio and greater information transfer rate compared with other BCI techniques. In this paper, a multiple signal classification based method was proposed for multi-dimensional SSVEP feature extraction. 2-second data epochs from four electrodes achieved excellent accuracy rates including idle state detection. In some asynchronous mode experiments, the recognition accuracy reached up to 100%. The experimental results showed that the proposed method attained good frequency resolution. In most situations, the recognition accuracy was higher than canonical correlation analysis, which is a typical method for multi-channel SSVEP signal processing. Also, a virtual keyboard was successfully controlled by different subjects in an unshielded environment, which proved the feasibility of the proposed method for multi-dimensional SSVEP signal processing in practical applications.

A real-time computational model for estimating kinematics of ankle ligaments.

BACKGROUND: An accurate assessment of ankle ligament kinematics is crucial in understanding the injury mechanisms and can help to improve the treatment of an injured ankle, especially when used in conjunction with robot-assisted therapy. A number of computational models have been developed and validated for assessing the kinematics of ankle ligaments. However, few of them can do real-time assessment to allow for an input into robotic rehabilitation programs. METHOD: An ankle computational model was proposed and validated to quantify the kinematics of ankle ligaments as the foot moves in real-time. This model consists of three bone segments with three rotational degrees of freedom (DOFs) and 12 ankle ligaments. This model uses inputs for three position variables that can be measured from sensors in many ankle robotic devices that detect postures within the foot-ankle environment and outputs the kinematics of ankle ligaments. Validation of this model in terms of ligament length and strain was conducted by comparing it with published data on cadaver anatomy and magnetic resonance imaging. RESULTS: The model based on ligament lengths and strains is in concurrence with those from the published studies but is sensitive to ligament attachment positions. CONCLUSIONS: This ankle computational model has the potential to be used in robot-assisted therapy for real-time assessment of ligament kinematics. The results provide information regarding the quantification of kinematics associated with ankle ligaments related to the disability level and can be used for optimizing the robotic training trajectory.

Assessment of movement quality in robot- assisted upper limb rehabilitation after stroke: a review.

: Studies of stroke patients undergoing robot-assisted rehabilitation have revealed various kinematic parameters describing movement quality of the upper limb. However, due to the different level of stroke impairment and different assessment criteria and interventions, the evaluation of the effectiveness of rehabilitation program is undermined. This paper presents a systematic review of kinematic assessments of movement quality of the upper limb and identifies the suitable parameters describing impairments in stroke patients. A total of 41 different clinical and pilot studies on different phases of stroke recovery utilizing kinematic parameters are evaluated. Kinematic parameters describing movement accuracy are mostly reported for chronic patients with statistically significant outcomes and correlate strongly with clinical assessments. Meanwhile, parameters describing feed-forward sensorimotor control are the most frequently reported in studies on sub-acute patients with significant outcomes albeit without correlation to any clinical assessments. However, lack of measures in coordinated movement and proximal component of upper limb enunciate the difficulties to distinguish the exploitation of joint redundancies exhibited by stroke patients in completing the movement. A further study on overall measures of coordinated movement is recommended.

Control strategies for effective robot assisted gait rehabilitation: the state of art and future prospects.

A large number of gait rehabilitation robots, together with a variety of control strategies, have been developed and evaluated during the last decade. Initially, control strategies applied to rehabilitation robots were adapted from those applied to traditional industrial robots. However, these strategies cannot optimise effectiveness of gait rehabilitation. As a result, researchers have been investigating control strategies tailored for the needs of rehabilitation. Among these control strategies, assisted-as-needed (AAN) control is one of the most popular research topics in this field. AAN training strategies have gained the theoretical and practical evidence based backup from motor learning principles and clinical studies. Various approaches to AAN training have been proposed and investigated by research groups all around the world. This article presents a review on control algorithms of gait rehabilitation robots to summarise related knowledge and investigate potential trends of development. There are existing review papers on control strategies of rehabilitation robots. The review by Marchal-Crespo and Reinkensmeyer (2009) had a broad cover of control strategies of all kinds of rehabilitation robots. Hussain et al. (2011) had specifically focused on treadmill gait training robots and covered a limited number of control implementations on them. This review article encompasses more detailed information on control strategies for robot assisted gait rehabilitation, but is not limited to treadmill based training. It also investigates the potential to further develop assist-as-needed gait training based on assessments of patients' ability. In this paper, control strategies are generally divided into the trajectory tracking control and AAN control. The review covers these two basic categories, as well as other control algorithm and technologies derived from them, such as biofeedback control. Assessments on human gait ability are also included to investigate how to further develop implementations based on assist-as-needed concept. For the consideration of effectiveness, clinical studies on robotic gait rehabilitation are reviewed and analysed from the viewpoint of control algorithm.

An intrinsically compliant robotic orthosis for treadmill training.

A new intrinsically compliant robotic orthosis powered by pneumatic muscle actuators (PMA) was developed for treadmill training of neurologically impaired subjects. The robotic orthosis has hip and knee sagittal plane rotations actuated by antagonistic configuration of PMA. The orthosis has passive mechanisms to allow vertical and lateral translations of the trunk and a passive hip abduction/adduction joint. A foot lifter having a passive spring mechanism was used to ensure sufficient foot clearance during swing phase. A trajectory tracking controller was implemented to evaluate the performance of the robotic orthosis on a healthy subject. The results show that the robotic orthosis is able to perform the treadmill training task by providing sufficient torques to achieve physiological gait patterns and a realistic stepping experience. The orthosis is a new addition to the rapidly advancing field of robotic orthoses for treadmill training.