Advances in additive manufacturing for medical robotics: A review

This paper proposes a modular variable configuration rehabilitation robot (MVCRR), aiming to meet the needs of multi-functional, full-cycle rehabilitation. MVCRR integrates a lower-limb training module and a-sit-to-stand module, offering 16 actuated degrees of freedom and supporting four rehabilitation postures: supine, sitting, standing, and sit-to-stand transition, while integrating the functions of typical rehabilitation devices. Meanwhile, a split–reassembly mechanism is designed to enable flexible bedside deployment, while compact scissor mechanisms and optimized actuation design improve adaptability and efficiency. In addition, a distributed bilateral coordination control architecture is established, and a unified human–robot interaction control framework is developed based on a multi posture dynamic model library, enabling compliant, assist-as needed control. Eight active and passive training modes are designed, covering gait reconstruction, muscle strengthening, posture transitions, and bilateral coordination. Experimental results demonstrate high-accuracy joint tracking (RMSE ≤ 0.0083 rad) and torque regulation (RMSE ≤ 1.6848 Nm), and good natural-gait reproduction (joint-angle RMSE ≤ 0.094 rad, r ≥ 0.897), validating the effectiveness of rehabilitation training across multiple postures. Overall, MVCRR establishes a systematic and intelligent integrated solution that supports multi-posture training and bedside deployment, demonstrating strong potential for clinical application.

Adaptive iterative learning control for upper limb rehabilitation robots considering unknown disturbance

Design and performance analysis of flexible wrist and elbow joint actuators for upper limb exoskeleton rehabilitation robot

ABSTRACT Background Stroke remains a leading cause of long‐term disability worldwide, and rehabilitation is essential for recovery. Although artificial intelligence (AI)‐related technologies have received growing attention in stroke rehabilitation, the knowledge structure and thematic evolution of this interdisciplinary field remain unclear. Objective To conduct a bibliometric analysis of AI‐related research in stroke rehabilitation from 2005 to 2024 and map publication trends, major contributors, thematic clusters, and emerging topics. Methods Relevant publications were retrieved from the Web of Science Core Collection (WoSCC), including SCI‐Expanded and SSCI, on November 30, 2024. Only English‐language articles and review articles published between January 1, 2005, and November 30, 2024 were included. A total of 3436 records were analyzed using CiteSpace 6.4.R1 Basic, GraphPad Prism 10.1.2, and biblioshiny in R. Analyses covered publication trends, collaboration networks, journal distribution, keyword co‐occurrence, clustering, and burst detection. Results Publication output increased markedly over time, with the United States contributing the largest number of publications. The Swiss Federal Institutes of Technology Domain was among the leading institutions, and Rocco Salvatore Calabrò was among the most productive and highly cited authors. Core publication venues included the Journal of NeuroEngineering and Rehabilitation and IEEE Transactions on Neural Systems and Rehabilitation Engineering . The literature mainly focused on virtual reality, upper‐limb rehabilitation, rehabilitation robotics, machine learning, cognitive rehabilitation, and transcranial direct current stimulation. Recent burst terms, including machine learning, artificial intelligence, and deep learning, indicated growing attention to data‐driven rehabilitation approaches. Conclusions AI‐related research in stroke rehabilitation has expanded substantially, with increasing emphasis on adaptive, data‐driven, and technology‐assisted approaches. This study provides a descriptive overview of the field's major trajectories, emerging gaps, and interdisciplinary directions, and may help inform future research and translational exploration.

The past, present and future of control architectures in lower-limb cable-driven robots for gait rehabilitation

Gait rehabilitation robots present promising solutions for delivering tailored training while optimizing clinical resources. This study presents an OpenSim simulation of the Mech-Walker, an innovative single-degree-of-freedom (DOF) gait rehabilitation device. By integrating the Mech-Walker with a detailed musculoskeletal model, we conduct a comprehensive investigation of walking dynamics and human-robot interaction. The device utilizes an 8-bar Jansen mechanism actuated by a single motor, enabling precise control of lower-limb motion, and incorporates an adaptable weight suspension system. Our simulation provides a detailed analysis of the interaction between the Mech-Walker and the human user, with particular focus on forces and torques at contact points and joints. Results indicate that the primary interaction forces at the pelvis-seat and foot-pedal interfaces are directly proportional to the user's body weight, with a peak vertical seat force of approximately 750 N for a 75 kg user. The mechanism effectively replicated hip and knee joint kinematics within normative ranges, while ankle trajectory exhibited greater deviation. This study establishes a validated simulation framework to support the design and refinement of gait rehabilitation robots.

With the rapid advancement of exoskeletons and rehabilitation robotics, modern healthcare increasingly demands high dynamic accuracy and reliability from medical devices. However, the dynamic response and durability of mechanical systems are greatly influenced by the inevitable existence of clearances in kinematic joints. Existing studies predominantly focus on simplified planar or spatial mechanisms, offering limited guidance for complex mechanical structures in medical applications. To address this issue, a unified modeling framework is proposed in this study to explore the nonlinear dynamic behavior and wear properties of bionic humanoid rigid mechanisms incorporating revolute joint clearances. A dynamic model that accounts for revolute joint clearances is established, employing the Lankarani–Nikravesh contact model alongside a refined Coulomb friction approach to characterize contact behavior. To characterize the wear progression between the shaft and the bushing, the Archard wear model is employed, while the system’s dynamic equations are formulated using the Lagrange multiplier approach. Systematic simulations are conducted to examine the effects of clearance size, location, and multi-clearance coupling on dynamic response and wear behavior. The results reveal that clearances at the hip joint have the most pronounced impact on system performance, tibiofemoral joint clearances exacerbate precision disturbances, and foot-end clearances considerably undermine system robustness. Increased clearance sizes and the coexistence of multiple clearances aggravate wear and induce more severe nonlinear dynamic phenomena. Phase portraits and Poincaré maps further illustrate that the system may exhibit complex or chaotic behavior under certain conditions. This study offers theoretical insights into performance degradation mechanisms in humanoid robots with joint clearances and introduces a modular “driving–mid–terminal” structure that enhances model generality, enabling its application to exoskeletons and rehabilitation devices for design optimization, service life prediction, and health monitoring.

Effects of robotic hand-assisted rehabilitation on motor function and daily living activities in acute stroke: a randomized controlled trial.

Regenerative medicine for stroke patients has been attracting attention. However, the effects of rehabilitation after the cell transplantation have not been fully elucidated. The purpose of the present study was to investigate whether intensive gait-focused rehabilitation using a robotic orthosis after regenerative medicine improved gait function and induced plastic changes in cortical networks. The present study was conducted in a retrospective cohort study. We selected seven chronic stroke patients, those who had undergone adipose-derived mesenchymal stem cells (MSC) transplantation therapy after the onset of stroke and had been receiving adequate subsequent gait rehabilitation with a robot for more than 2 months. During hospitalization, each patient received at least 2 h of rehabilitation, including robotic-assisted gait training more than five times per week. As the assessments, gait performance and M1 seed-based resting state-functional connectivity (rs-FC) obtained by a magnetoencephalography were compared before and after hospitalization. After rehabilitation, cadence and spatial gait symmetry ratio were significantly improved, and a significant negative correlation was found between the changes in the gait symmetry ratio and the time from transplant to rehabilitation. Seed-based rs-FC in the beta band between the lesioned M1 and multiple brain regions (e.g., both frontal areas, ipsilateral postcentral gyrus) was significantly decreased after the rehabilitation. Significant negative correlations were also observed between the changes in the gait symmetry ratio and the changes in lesioned M1 seed-based rs-FC in the paracentral gyrus and regions associated with the default mode network. It was revealed that intensive gait-focused rehabilitation using a robotic orthosis improved gait function and induced plastic changes in the cortical networks. The improvements were significantly correlated with the timing of the start of rehabilitation after MSC transplantation.

Neurorehabilitation in physiotherapy depends on experience-dependent neuroplasticity; however, conventional clinical outcomes may lack sensitivity to capture dynamic neural adaptations underlying recovery. Brain entropy and complexity measures derived from EEG and neuroimaging have emerged as potential biomarkers of neural adaptability. To map and synthesize evidence on brain entropy and complexity as biomarkers of neuroplasticity in neurorehabilitation, with relevance to physiotherapy practice. A scoping review was conducted following PRISMA-ScR guidelines. PubMed, Scopus, and Web of Science were searched up to August 2025 for studies reporting quantitative entropy or complexity measures in neurological populations undergoing rehabilitation or task-based assessment. Eight studies were included. Interventional studies in stroke and brain injury populations reported moderate to large within-group neural effects, with increases in entropy or complexity accompanying functional improvement following task-oriented, robotic, or brain-computer interface-based rehabilitation. Studies of higher methodological quality demonstrated more consistent entropy-outcome associations, whereas lower-quality observational studies showed greater variability. Degenerative neurological conditions are characterized by reduced neural complexity. Brain entropy and complexity measures are sensitive indicators of neuroplastic change and may complement clinical outcomes in physiotherapy. Although not yet ready for routine clinical decision-making, these biomarkers show promise for monitoring intervention response and guiding personalized rehabilitation, pending methodological standardization and longitudinal validation.

This article presents a novel knee exoskeleton actuator based on a planar rotary spring, aimed at improving torque transparency and reducing structural complexity in wearable robotic applications. To achieve a compact and lightweight design, the spiral geometry of the neutral surface and arm thickness is optimized using a physics-based model under multiple mechanical constraints. To enable compliant control, a cascaded impedance controller was implemented, with feedforward and friction compensation integrated into the inner loop torque control. To improve torque transparency, the effect of a disturbance observer with a leveraging coefficient was analyzed and compared under different damping ratios. Passivity is maintained through adaptive gain regulation and a velocity-threshold-based time domain passivity approach. The resulting prototype achieves a high torque-to-mass ratio compared to existing series elastic actuator designs. Experimental results further confirm the system’s robustness, safety, and torque transparency. In particular, the proposed method reduces the residual torque to 0.62 N<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\cdot$</tex-math></inline-formula>m at 1 Hz, demonstrating improved performance compared to conventional proportional–derivative control. In addition, the actuator achieves accurate impedance rendering during dynamic interaction tasks, highlighting its potential for wearable robotic assistance and rehabilitation applications.

Hip-knee coordination control and clinical validation of a horizontal lower-limb rehabilitation robot based on human-machine coupling dynamics modeling

Reinforcement learning-based assist-as-needed control for shoulder rehabilitation robot based on virtual biomechanical model

Rehabilitation robotics faces the challenges of aligning engineering design with patient-specific needs. Most existing controllers in rehabilitation robots often constrain motion to fixed paths or provide only passive guidance, limiting user engagement and adaptability. This study proposes a novel active-assistive mode controller that integrates a virtual tunnel-based force generation mechanism with a dynamic moving-window technique for tracking activities of daily living (ADL) trajectories. Unlike conventional impedance controllers, the proposed method dynamically adjusts the virtual tunnel in real time, permitting voluntary upper-limb movement within a safe operational range while preventing excessive deviation. The system was implemented on a wearable two-degree-of-freedom (DOF) upper-limb exoskeleton equipped with drive and integrated sensor units. Experimental results demonstrated that decreasing the guidance force (Fgf) increased tracking errors, from 1° at 100% Fgf to 5° at 30% Fgf, indicating greater voluntary participant motion. Peak actuator torques correspondingly decreased from 14.75 to 13.43 Nm (elbow) and from 4.14 to 2.48 Nm (wrist), confirming the controller's capability to modulate robotic assistance according to user effort. Tests with 30 healthy participants demonstrated the effectiveness of guidance along predefined ADL trajectories, validating the controller's potential for patient-centered rehabilitation.

Dear Editor, We read the interesting letter to the Editor by Dr. Finsterer and appreciated the comments on our work. In this paper we aim at answering to the questions raised. For the first point, Dr Finsterer suggests to add information about Nerve Conduction Studies (NCS) and needle Electromyography (EMG); unfortunately, we are not able to give the neurophysiological investigations acquired during the first admission at the hospital. Anyway, as regards the type of GBS, we can point out that he was diagnosed with an Acute Motor Axonal Neuropathy (AMAN), as stated by previous medical documents. Furthermore, we can confirm that the patient presented cranial nerve involvement, as evidenced by the ascending paralysis which involved trunk and head control, swallowing problems, mimic musculature deficits and respiratory impairment.[...].

This study examines human arm kinematics during overground physical human-robot interaction (pHRI). Previous work showed humans adjust arm stiffness with changing trajectory uncertainty, but the roles of arm kinematics and muscle activation remained unclear. Building on a preliminary study, we analyzed arm movements with more participants (10 individuals) to achieve more reliable findings and examined two potential influences that arose in the preliminary study: the robot’s perturbation effect (a brief hand push) and left-to-right trunk sway. Using a linear mixed-effects model, we evaluated the effects of participant, block, and trajectory condition on arm angles. Results showed minimal kinematic contribution to stiffness modulation, with inconsistent significance levels in the measured metrics. Perturbation presence also had no significant impact on voluntary posture, with the exception of the posture differences before and after the perturbation. Trunk sway was strongly correlated with elbow angle, with a mean correlation (R2) of 0.65 and a standard deviation of 0.24. Most variability arose from individual differences rather than experimental conditions. These findings might potentially allow for more flexible mechanical design in assistive and rehabilitation robots.

Adaptive control of lower-limb rehabilitation robots is challenging due to the nonlinear dynamics of coupled joints, patient-specific uncertainties, and external disturbances during human–robot interaction. This study presents a hybrid quantum-inspired adaptive intelligent fuzzy control framework for a 3-DOF lower-limb rehabilitation robot. The quantum approximate optimization algorithm (QAOA) is used to tune fuzzy membership function parameters online, while the fuzzy logic controller uses joint position error and its derivative in a closed-loop configuration. Triangular membership functions are adjusted in real time to minimize the integral of squared error. The proposed controller is evaluated through simulations on a nonlinear dynamic model of the robot, including coupled joint interactions and typical disturbances. The results indicate the improvements in trajectory tracking accuracy, disturbance rejection, and energy dissipation compared to conventional fuzzy logic controllers and recent adaptive or reinforcement-learning-based methods. An energy-based analysis combined with Lyapunov stability assessment to confirm the enhanced closed-loop stability, showing faster energy dissipation for the QAOA-optimized fuzzy controller. The results ensure that integrating quantum optimization with fuzzy control can improve robustness and accuracy of rehabilitation robots, effectively handling nonlinearities and patient-specific uncertainties. The study employed Hardware-in-the-Loop testing to prove that the proposed controller achieves its intended performance during actual operation although the system requires ongoing monitoring for any performance-related issues. Results demonstrate that the QAOA-optimized fuzzy controller reduces the integral squared error (ISE) by approximately 96–99%, settling time by 62–63%, and limiting peak overshoot by nearly 75–80% across all considered test systems relative to previously reported controllers.

Effects of Rehabilitation Robot Training on Cognitive Function in Stroke Patients: A Meta-Analysis.

Abstract Paediatric orthopaedic disorders encompass a broad and heterogeneous spectrum of congenital, developmental, infectious, neoplastic, and traumatic musculoskeletal conditions that selectively affect the immature, dynamically growing skeleton. The physiological distinctiveness of the paediatric musculoskeletal system characterised by open physes, active longitudinal bone growth, remarkable remodelling capacity, and age-dependent biomechanical properties necessitates diagnostic frameworks and therapeutic strategies fundamentally different from those applied in adult orthopaedics. The discipline has undergone a substantial paradigm shift over recent decades, transitioning from descriptive, radiograph-centred clinical assessment towards technologically enhanced, evidence-based, and multidisciplinary models of care. The present study synthesises contemporary diagnostic and therapeutic developments in paediatric orthopaedics and presents an empirical analysis of functional recovery predictors drawn from a hospital-based sample of 360 paediatric patients managed in a tertiary care unit. Using one-way analysis of variance (ANOVA) and multiple linear regression modelling, the study identifies and quantifies the principal determinants of postoperative and rehabilitative functional improvement. Advanced imaging integration incorporating high-resolution magnetic resonance imaging (MRI), three-dimensional computed tomography (CT) reconstruction, and intraoperative navigation emerged as the strongest independent predictor of functional recovery (β = 0.44, p &lt; .001). Early referral to specialist paediatric orthopaedic services was the second most influential predictor (β = 0.36, p &lt; .001), reinforcing the critical importance of timely diagnosis during periods of active skeletal growth. Multidisciplinary care coordination among orthopaedic surgeons, paediatricians, physiotherapists, and rehabilitation specialists produced independent positive effects on outcome trajectories (β = 0.31, p &lt; .01). Conversely, delayed clinical presentation was associated with significantly elevated complication rates and reduced functional scores (β = −0.39, p &lt; .001). The integrated regression model accounted for 71% of the variance in functional recovery outcomes (R² = 0.71, F(4, 355) = 219.63, p &lt; .001), affirming its strong explanatory power and clinical relevance. These findings collectively corroborate the contemporary evidence base supporting early diagnosis, minimally invasive and growth-preserving surgical strategies, integrated postoperative rehabilitation, and the progressive incorporation of precision medicine, biological therapies, rehabilitation robotics, AI-assisted imaging interpretation, and digital health monitoring within child-centred orthopaedic care frameworks. Keywords: paediatric orthopaedics; diagnostic imaging; musculoskeletal disorders; minimally invasive surgery; paediatric trauma; multidisciplinary care; precision medicine; functional recovery; growth plate; rehabilitation