OBJECTIVES:This study aims to investigate the clinical effect of acupuncture combined with rehabilitation training in the treatment of knee osteoarthritis (KOA) under unified acupoint intervention. PATIENTS AND METHODS:In this prospective, comparative study informed by systematic acupoint analysis, the literature related to acupuncture for KOA treatment was retrieved from public databases and the core acupoints were analyzed using the Traditional Chinese Medicine Inheritance Calculation System (TCMICS) software. Eighty-four patients were selected as the participants, and the clinical effect (total efficacy rate), Numeric Rating Scale (NRS) for pain score, American Knee Society Score (AKSS) for knee joint function and the self-care ability score for daily living activities were used to evaluate the treatment effect. The curative effect index was calculated using the AKSS scores. RESULTS:After analyzing the core acupoints, we selected the Dubi, Zusanli, Yanglingquan, Xuehai, Neixiyan and Liang Qiu acupoints for the acupuncture treatment. The patients were divided into a control group (n = 42 patients undergoing rehabilitation training) and an intervention group (n = 42 patients undergoing acupuncture combined with rehabilitation training). The control group consisted of 15 male and 27 female, with a mean age of 62.90 ± 11.531 (range, 40 to 78) years. The mean course of the disease was 4.07 ± 1.841 years. The intervention group consisted of 16 male and 26 female, with a mean age of 63.95 ± 10.613 (range, 41 to 79) years and a mean disease course of 4.76 ± 2.328 years. There was no significant difference between the groups before treatment (p > 0.05). After acupuncture treatment, the 92.85% total efficacy rate (n = 39) was significantly higher than the 73.80% (n = 31) in the control group. The mean AKSS of the control and intervention groups were 64.17 ± 11.148 and 66.67 ± 17.795, respectively, and the mean knee joint articulation scores were 57.40 ± 8.925 and 54.10 ± 10.150 in the two groups, respectively. Following treatment, the scores of knee joint function and knee joint articulation in the intervention group were higher than those in the control group (p < 0.05). For AKSS, the mean difference of the control group was 6.030, and the proportion of patients who reached the improvement of minimum clinically significant difference (MCID) standard was 80.95%. The mean difference of the intervention group was 10.98, and the proportion of patients who reached MCID was 100%. Following treatment, the self-care ability score in the intervention group were higher than those in the control group (p < 0.05). For modified Barthel Index (MBI), the mean difference of the control group and intervention group were 6.446 and 6.225, all accounting for 100%. The mean NRS scores were significantly decreased from 5.19 ± 1.042 to 1.93 ± 0.640 after one month of acupuncture treatment. CONCLUSION:Our study results suggest that acupuncture combined with rehabilitation training based on data mining can relieve patients' pain, accelerate the recovery of their knee joint function and improve their daily living ability, demonstrating high clinical value in practice.

Assistive bathing for the elderly and disabled presents significant challenges regarding caregiver workload and safety. This paper presents the design and verification of a multi-module integrated intelligent bathing assistance system. The system automates the entire bathing sequence through four coordinated modules: a robotic scrubbing unit, a climate-controlled cabin, a passive multifunctional wheelchair, and a multi-degree-of-freedom transfer device. A key innovation is the wheelchair’s passive design with an automated docking mechanism, ensuring safety in wet environments. Unlike existing commercial solutions and the existing literature, which primarily focus on fragmented, singular functionalities (such as transfer-only devices or fixed-spray cabins), the core advantage of the developed system lies in its holistic integration of safe physical transfer, adaptive robotic scrubbing, and microenvironment control into a seamless, unified architecture. Employing a modular and ergonomic approach, the system executes a predefined 12-step automated workflow. Experimental validation demonstrates an average bathing time of 16.6 min and a quantifiable 69.8% reduction in caregiver workload, confirming the system’s high efficiency and practical utility in alleviating caregiver burden.

INTRODUCTION:With the increasing requirements for performance and safety of lower limb exoskeleton rehabilitation robots, researchers are exploring various active compliance control strategies to improve the compliance of control and enhance safety and effectiveness of rehabilitation training. The focus of this paper is to evaluate existing active compliance control strategies applied to lower limb exoskeleton rehabilitation robots, highlighting their principles, key technologies, and performance. AREAS COVERED:Literature searches were conducted in PubMed and Web of Science from 2000 to 2025, starting with the broad keyword 'active compliance control strategy' and narrowing down to specific aspects such as 'impedance control'. Sixty-nine articles were reviewed, and eight active compliance control strategies were divided into three categories according to their core principles: impedance control (n = 18), force, and position hybrid control (n = 27) and machine learning-based hybrid control (n = 24). Their applications in five types of lower limb exoskeleton rehabilitation robots are outlined. EXPERT OPINION:Active compliance control strategies for lower limb exoskeleton rehabilitation robots are necessary for safe human-robot interaction and assist-as-needed implementation. This paper envisions the concept of an active compliance controller that covers the entire rehabilitation cycle, ensuring that the robot is continuously aligned with the patient's rehabilitation needs.

Shape memory alloys (SMAs) are a class of smart materials exhibiting unique superelastic behavior, making them highly attractive for biomedical applications. Constitutive modeling plays a crucial role in bridging the intrinsic material response and engineering design. In this work, the development of superelastic constitutive models for SMAs is systematically reviewed from both macroscopic and microscopic perspectives. Macroscopic phenomenological models provide computationally efficient tools for engineering applications by simplifying phase transformation mechanisms, while microscopic models capture the underlying crystallographic evolution and offer deeper insights into hysteresis behavior and functional degradation. In addition, recent advances in experimental characterization, including tension–compression cycling, fatigue loading, and environmental testing, are summarized to highlight the key factors influencing superelastic performance and durability. The integration of constitutive models with finite element analysis enables accurate prediction of stress distribution, phase transformation evolution, and long-term stability of complex biomedical devices under physiological conditions. Finally, the role of superelastic SMAs in biomedical applications, such as stents and orthopedic implants, is discussed, and future research directions are outlined. This review provides a comprehensive foundation for the design and optimization of SMA-based biomedical devices.

Wearable lower-limb prostheses benefit from environment-aware control, yet strict on-device constraints (size and compute) make real-time perception and decision-making challenging. We present an edge-computing framework that integrates two inertial measurement units with an event-triggered array LiDAR sensor to enable synchronous locomotion modes prediction and terrain features calculation. Joint kinematics from the IMUs dynamically trigger LiDAR scans at critical gait phases, reducing LiDAR active sensing time while preserving forward terrain information for decision-making. The complete pipeline is deployed on an embedded platform with deterministic real-time operation. In experiments with six healthy participants and two hip-disarticulation amputees across five terrains in indoor and outdoor routes, the system achieved an overall locomotion-mode prediction accuracy of 97.85% (98.80% indoor, 96.90% outdoor) and 4.87% mean relative error for terrain features calculation. The end-to-end decision latency was under 11.5 ms, supporting low-latency prosthetic control.

Intelligent prosthetic hips and knees represent a critical advancement in restoring natural gait and mobility for lower-limb amputees, particularly those with high-level amputations such as hip disarticulation. This systematic review examines recent progress in three fundamental aspects of intelligent prosthetic technology: actuation, perception, and control. In terms of actuation, the review highlights the limitations of passive and active prostheses and discusses emerging hybrid active-passive mechanisms that aim to replicate the natural, biarticular muscle-driven energy transfer in human gait. The perception section addresses current methodologies for recognizing human motion intentions through mechanical, bioelectric, biomechanical, and external environmental signals, underscoring the challenges of stability, latency, and interference inherent in existing approaches. Regarding control strategies, the paper categorizes intelligent control into torque compensation, motion following, and direct intention control, outlining the strengths and limitations of each method. The review identifies critical technological bottlenecks, including signal interference, limited adaptability to dynamic environments, and the absence of effective real-time intention recognition methods. The paper concludes by suggesting future directions in the development of hybrid actuation and advanced perception-control integration, essential for improving the usability and efficacy of intelligent prosthetic hips and knees, ultimately enhancing mobility and quality of life for amputees.

Lower-limb exoskeletons require lightweight terrain perception to support terrain-adaptive locomotion assistance. Existing methods mainly focus on coarse terrain classification, while fine-grained terrain parameters such as stair height are less explored. This paper proposes a lightweight two-stage hierarchical decoupled terrain recognition framework for lower-limb exoskeletons using a single VL53L5CX depth sensor. In the first stage, a CNN-based classifier performs coarse terrain recognition among level ground, stair ascent, stair descent, and others. In the second stage, a CNN-based regressor further estimates stair height for stair-ascent scenes. Field experiments show that the first-stage network achieves an overall terrain-recognition accuracy of 96.9%. The second-stage network achieves a stairheight prediction MAE of 0.47 cm, outperforming an MLP baseline with a MAE of 0.62 cm. The proposed method can run on an STM32F407 microcontroller, demonstrating its feasibility for lightweight embedded terrain perception in lower-limb exoskeletons.

A tendon-driven hybrid rigid-flexible manipulator with orthogonal hinges is presented for precision spraying applications in unstructured environments. The proposed architecture integrates waterproof electronics and simplified cable routing to enhance environmental adaptability while reducing system inertia. The primary contribution of this work is an integrated mechanical-algorithmic framework. Within this framework, we adapt a chaos-enhanced Particle Swarm Optimization (CE-PSO) algorithm, utilizing established chaotic initialization and adaptive parameter adjustment techniques, to resolve redundant inverse kinematics with 1.28% mean positioning error. This solver is constrained by a practical hierarchical classification strategy that mitigates joint coupling effects. Standard fuzzy PD control and quintic trajectory planning were employed solely to facilitate the validation of this mechanical-kinematic framework, enabling ±30° bending with under 2.39° joint coordination error. Experimental validation confirms a computation time of 0.85s for the inverse kinematics solution, which is suitable for quasi-static or slow-motion tasks such as precision spraying, but may be limiting for highly dynamic operations requiring faster update rates. Notably, dynamic operations exhibit ≤4.65mm end-effector oscillations due to structural compliance. Collectively, the integrated mechanical-computational framework provides a hardware and algorithmic foundation for precision tasks in less structured settings. While kinematic accuracy and static stability were validated in lab tests, comprehensive task-level evaluations and formal safety characterizations are required to fully realize its potential for safe human-robot interaction. However, further real-world validation is necessary to fully assess task-level performance such as spray coverage uniformity and human-robot interaction safety.

This study introduces a novel auxetic metamaterial structure specifically engineered for protective sports equipment through a parametric design and additive manufacturing approach. Drawing inspiration from the intricate patterns of traditional Persian Lori rugs, a reentrant tubular lattice is conceived as a three-dimensional metamaterial capable of exhibiting a tunable negative Poisson's ratio. The structure is fabricated using high-resolution digital light processing (DLP) 3D printing with an ABS-like photopolymer, enabling precise reproduction of the complex geometry. Systematic variation of two key geometric parameters, wall thickness (0.8, 1.0, 1.2 mm) and cell width (2.75, 4.0, 5.25 mm), allowed rigorous parametric control of mechanical behavior. Combined finite-element analysis and experimental compression testing verified exceptional tunability in stiffness, energy absorption, and Poisson's ratio, which ranged from -1.09 to -2.3. The configuration with 1.2 mm thickness and 5.25 mm width demonstrated the highest stiffness and impact-energy absorption, highlighting its potential for helmets, elbow pads, and similar high-impact gear. The integration of culturally inspired geometry, metamaterial design principles, and precision DLP 3D printing establishes a unique pathway for next-generation protective equipment, showcasing how parametric control of auxetic metamaterials can simultaneously achieve lightweight construction, superior energy dissipation, and enhanced user comfort.

Surface electromyography (sEMG) provides a non-invasive measure of the neural drive transmitted from the central nervous system to muscles by capturing the spatiotemporal summation of motor unit action potentials at the skin surface, and is therefore widely used to study neuromuscular coordination during motor tasks. By reflecting neural drive transmitted from the central nervous system to peripheral muscles, sEMG provides valuable insights for investigating neuromuscular coordination during upper-limb motor tasks. Within the framework of modular motor control, muscle synergy analysis has been increasingly applied to characterize coordinated muscle activation patterns extracted from multi-channel sEMG recordings. In this study, sEMG signals were collected from twelve stroke patients and nine healthy subjects during robot-assisted upper-limb training, involving two movement trajectories (straight and rectangular) and multiple robot-assisted levels. Muscle synergies were extracted using non-negative matrix factorization (NMF). A synergy merging-splitting model, combined with a Functional Driving Ratio (FDR), was employed to characterize both the muscle synergy reorganization and the relative activation contributions of driving versus stabilizing muscle components in terms of motor control strategy. The results showed that healthy subjects maintained consistent muscle coordination patterns across different assistive levels, while making task-dependent adjustments to muscle activation to adapt to variations in movement trajectories. For stroke patients, higher functional status was correlated with more differentiated coordination patterns and relatively higher FDR values, suggesting greater reliance on task-relevant agonist muscles during movement execution. In contrast, lower-function patients exhibited less differentiated coordination patterns accompanied by reduced FDR values, indicating the increased involvement of stabilizing or antagonist muscles. This shift may reflect compensatory control strategies and the reduced efficiency of neuromuscular coordination during assisted upper-limb movements. These findings suggest that sEMG-based muscle synergy features and the FDR may provide quantitative, sensor-derived support for characterizing neuromuscular coordination during robot-assisted rehabilitation.

Background: The shoulder joint is a pivotal articulation for upper-limb kinematics, yet it is prone to injury during rehabilitation exercises. Despite advances in assistive technologies, current rehabilitation exoskeletons still face challenges in reducing the risk of glenohumeral subluxation while also improving load-bearing capacity. Methods: This work proposes a novel Pneumatic Shoulder Exosuit (PSE), designed to protect the shoulder joint from subluxation while augmenting its load-bearing capability during rehabilitation. This dual functionality is enabled by two coupled, pouch-based pneumatic actuators configured into a bionic agonist/antagonist pair. Device performance was evaluated through experimental trials with 10 volunteers performing kinematic assessment, standardized trajectory-tracking, and static overhead weight-holding tasks. In addition, 7 participants with post-stroke hemiparesis were involved in evaluating the satisfaction with Activities of Daily Living (ADL) rehabilitation training using the device. Results: Experimental results showed that the PSE can provide dynamic gravity support with up to 24 Nm of torque. The device maintained the movement trajectory with a 96.5% similarity compared to unassisted. It also reduced agonist muscle activities (anterior, middle, and posterior deltoids and biceps brachii) by up to 65% with a 70% reduction in compensatory muscle (latissimus dorsi) activity, and a 45% reduction in peak lateral trunk tilt. The participants with post-stroke hemiparesis also provided positive feedback. Conclusions: This study demonstrates the strong potential of the PSE to effectively combine load-bearing augmentation with kinematic transparency, offering new insights into bionic soft actuator design. These findings suggest broad applicability in safe and effective rehabilitation across clinical and home settings. Future research should further explore adaptive assistance strategies during complex dynamic ADLs to refine the functional utility of such systems.

Surface electromyography (sEMG) offers an effective communication channel for human-machine interfaces (HMI), yet achieving robust myoelectric control across varying limb positions remains a significant challenge, particularly for amputee subjects. This study introduces a novel approach combining a Virtual-Real Fusion Target Achievement Control (VRF-TAC) system with a multi-day dynamic asymmetric training strategy to address these limitations. The VRF-TAC system, implemented with HoloLens 2, creates a context-informed virtual-real embodiment where immersive visual feedback on gesture and muscle contraction intensity partially supplants lost tactile sensation, thereby lowering the user learning barrier. Concurrently, the dynamic asymmetric training protocol intentionally introduces varied and challenging conditions to drive robust learning and facilitate the translation of control strategies to the personalized conditions of amputee subjects. The multimodal progressive domain adversarial neural network (MPDANN) algorithm was employed to handle inter-session and inter-position signal variations. Evaluation across ten hand gestures and five arm positions with both healthy and amputee subjects demonstrated statistically significant improvements (p < 0.01) in key performance metrics, including throughput, overshoot and path efficiency, with the average completion rate reaching 92.44% +/- 3.55% on the final day. Our findings validate that this integrated approach significantly enhances controller resilience to limb position changes and presents a promising strategy for extending co-adaptive benefits to amputee subjects.

Galt analysis is crucial for disease diagnosis and rehabilitation assessment; however, traditional optical motion capture systems are costly and limited to fixed setups. This study presents an "Offline Estimation Method for Hip and Knee Joint Angles of Lower Limbs Based on Quaternion and DTW Time Alignment." The method uses quaternion fusion of inertial measurement unit (IMU) orientations and employs Sakoe-Chiba constrained Dynamic Time Warping (DTW) to eliminate a 42 ms initial offset and a 150 ms cumulative drift. Combining N-pose calibration with heel velocity event detection allows for the offline calculation of hip and knee joint angles. Data were collected from eight healthy participants during flat walking and stair ascent/descent scenarios, with the Noraxon Ultium Motion system serving as the reference. Results show that DTW reduces the average root mean square error (RMSE) by 29.1%; specifically, "the RMSE for hip flexion reaches 4.1 degrees, while the overall knee joint RMSE is 10.2 degrees, with correlation coefficients >= 0.87. Hip joint measurements consistently met the clinically acceptable threshold of <10 degrees across all scenarios; knee joint measurements satisfied this threshold during flat walking (RMSE = 7.8 degrees) but exceeded it during stair negotiation (RMSE = 11.4 degrees), reflecting the increased biomechanical complexity of multi-planar knee motion during stair activities. This study provides a low-cost, high-precision solution for the post-hoc offline estimation of hip and knee joint angles. The proposed method is specifically designed for retrospective gait data analysis rather than real-time feedback, offering a scalable strategy for early screening of gait abnormalities and clinical assessment in home and community rehabilitation settings.

PurposeConstipation is a prevalent disorder of the digestive system, and abdominal massage is commonly employed to alleviate its symptoms due to its non-invasive nature and minimal side effects. This study aims to develop an accurate correlation model that links massage force (MF), massage depth (MD), and intra-abdominal wall deformation (IAWD) based on patient-specific data, thereby providing quantitative standards and personalized treatment options for the use of clinical massage devices.MethodsThe study utilized finite element method (FEM) and machine learning (ML) techniques to construct a digital twin model of the abdomen. Initially, a high-fidelity skin-muscle two-layer abdominal finite element model was developed using computed tomography (CT) images and mechanical experimental data. Subsequently, four individual learning models were trained, and stacked models were constructed by combining the advantages of these individual models through random forest and K-nearest neighbor algorithms to predict MF and IAWD.ResultsThe prediction accuracy rates for MF in the ascending colon, transverse colon, and descending colon regions were 84.80%, 84.00%, and 85.60%, respectively. The prediction accuracy rates for IAWD were 96.00%, 87.00%, and 95.20%, respectively.ConclusionCompared to traditional finite element simulations, this abdominal digital twin model is capable of delivering predictions within 3 seconds, offering valuable insights for the digital transformation of healthcare.

Objective: This paper presents a novel energy-based gait adjustment approach for adapting assistance to individuals with different gait abnormalities to improve gait performance. Method: This paper first proposes a Lower-Limb Bidirectional Potential Energy Transmission (BPET) model combined the human leg and an elastic energy storage element (EESE). Then, the stiffness of the EESE is modeled based on the BPET model. A simplification model is derived using the least-squares method to obtain a personalized constant stiffness K, which can be utilized for personalized gait assistance. Finally, the BPET model based gait assistance module prototype was manufactured and evaluated in gait-adjustment experiments involving six healthy subjects with simulated gait abnormalities and one subject with cerebral palsy (CP). Nine indices were adopted to characterize gait performance. Results: In healthy subjects, module was associated with increases in lower-limb kinematic and spatial gait parameters. Under the different loading conditions, mean increases were observed in peak hip flexion (16.8% ± 6.5% and 17.3% ± 16.2%), hip range of motion (ROM, 16.4% ± 14.8% and 19.3% ± 6.9%), toe clearance (45.0% ± 53.5% and 61.5% ± 62.8%), step length (21.0% ± 14.7% and 22.9% ± 8.0%), and stride length (16.1% ± 14.0% and 20.9% ± 12.5%). Peak knee flexion increased in five of six subjects under each loading condition. In the CP subject, module increased peak hip flexion, hip joint ROM, toe clearance, step length, stride length, and peak knee flexion by 46.8%, 27.6%, 32.3%, 9.6%, 6.0%, and 10.7%, respectively. Across all subjects, temporal symmetry indices shifted toward the ideal value under module. Conclusion & Significance: This work provides preliminary evidence that personalized elastic assistance, governed by a user-specific model, may improve gait performance during rollator-assisted walking.

In the process of assisting, following, and providing safety protection for users with walking difficulties, the active obstacle avoidance behaviour of the intelligent walker is of great importance. In light of this, a human-machine shared control strategy based on the feed-forward neural network artificial potential field method (NN-APF) is proposed, which is suitable for the local path planning requirements of Intelligent walker. The proposed methodology is founded on the NN-APF algorithm, which employs a neural network to calibrate the repulsive gain coefficients of the artificial potential field (APF), thus enhancing the conventional APF technique. This reduces the likelihood of the APF getting stuck in local minima or failing to reach the target, thereby enhancing the algorithm's stability. NN-APF was tested on the MATLAB simulation platform. The simulation experiments demonstrated that NN-APF possesses the capability to safely navigate obstacles, achieving a target point arrival rate of 92.3%. Furthermore, it has been shown to enhance the trajectory smoothness of Intelligent walker in complex environments. Furthermore, experimental trials on the intelligent walker demonstrated the feasibility and practicality of NN-APF. Following the deployment of NN-APF, the intelligent walker is capable of actively avoiding obstacles while adhering to the user's designated walking path and providing guidance to select a safer route.

A two-wheeled self-balancing wheelchair is a highly nonlinear and underactuated wheeled inverted-pendulum system, which faces severe balance control challenges under large tilt angles, load variations, and external disturbances. A hybrid nonlinear model predictive control-linear quadratic regulator (NMPC-LQR) balance control scheme is presented that combines nonlinear optimal recovery and efficient local stabilization while reducing the computational burden. A reduced-order nonlinear dynamic model of the wheelchair-occupant system is derived via the Lagrange method. The proposed controller confines LQR to a locally valid region verified by linearization-error analysis, whereas NMPC is employed for large-deviation recovery. A linearization-error analysis is first used to determine the switching-threshold search range, and a normalized five-index criterion identifies 0.100 rad as the optimal nominal threshold. Mass-sensitivity tests further provide an empirical relation for estimating the optimal threshold within the tested mass range. Comparative simulations under initial tilt angles of 35, 40, and 45 degrees show faster convergence than standalone LQR and standalone NMPC controllers. Benchmark comparisons with SMC demonstrate a larger recoverable initial-angle range. Robustness tests under parameter variations, pulse and step disturbances, and flat-ground friction changes, together with prototype experiments, validate the stability, disturbance rejection, and engineering applicability of the proposed control strategy.

In the rehabilitation training of stroke patients, traditional rigid-driven robots struggle to effectively address spasticity, leading to suboptimal training and risks of secondary injury. This paper proposes a dual-motor driven variable stiffness elbow Joint rehabilitation robot based on a cam-spring-roller mechanism, enabling independent control of stiffness and position. We developed an LSTM-based elbow joint torque estimation model trained on a dataset generated using a musculoskeletal model incorporating Hill muscle model, achieving accurate and reliable torque prediction. A stiffness and impedance cooperative control framework is then established based on the estimated torque values: wherein the system provides higher assistive torque with lower mechanical stiffness when lower muscle strength is detected, and reduces assistive torque while increasing mechanical stiffness when higher muscle strength is identified. The framework also includes a spasticity detection model is proposed, incorporating metrics of antagonist muscle abnormal activation, normalized jerk score and velocity-activation quantification, aiming to enhance training safety and demonstrating the potential for improved patient engagement. Finally, experiments on four healthy subjects and a patient with spasticity demonstrated the feasibility and preliminary efficacy of the stiffness-variable impedance control framework and the spasticity detection model in a pilot study.

This article designs a novel ankle rehabilitation based on the remote center-of-motion (RCM) mechanism, which can be adapt to various training scenarios such as sitting and lying-down training. In order to explore the impact of human soft tissue viscoelasticity on the comprehensive performance of the rehabilitation, the relationship between the human-rehabilitation interaction torque and the posture deviation of the rehabilitation are quantitatively analyzed and expressed by kinematic branch chains. Then, a coupling mathematical model of the human rehabilitation is further constructed according to the virtual equivalent parallel mechanism (VEPM) modeling method, and performance indicators (workspace, angle deviation, motion center deviation, condition number ratio, and motion/force transmission) of the VEPM model are derived. The performance of the VEPM under different interaction forces are analyzed and verified by simulation, motion experiments, and motion coordination experiment. The results of this article have positive significance for further exploring the human-robot physical interaction principle and improving the human-robot collaboration of wearable exoskeletons.

Passively adaptive robotic hands have attracted growing attention because they shift part of the burden of grasp adaptation from sensing and control to mechanical design. However, existing reviews often discuss variable stiffness, underactuation, structural deformation, and related strategies separately, making it difficult to compare how different passive mechanisms balance adaptability, load capacity, grasp precision, and control simplicity. This paper provides a trade-off-driven and application-oriented review of passively adaptive robotic hands. The field is synthesized from a mechanism-performance-application perspective, covering four major categories: variable-stiffness mechanisms, underactuated coupling and transmission mechanisms, structural deformation mechanisms, and mode-switching and hybrid actuation strategies. These mechanisms are discussed in relation to key performance indicators, including shape adaptability, payload, positional tolerance, impact resistance, lightweight design, and reliability, as well as their relevance to representative scenarios such as daily object grasping, high-load manipulation, and clinical assistive applications. Rather than maximizing any single performance metric, passive adaptive mechanisms primarily contribute by redistributing competing design constraints through mechanical intelligence. Finally, this paper highlights current challenges, discusses limitations in cross-study performance comparison, and outlines promising directions for future research.