Research Progress of Intelligent Intervention Modeling for Ankle

Injury Rehabilitation and Prevention

Ziyi Liao

Medical college, Hunan University Of Chinese Medicine, Changsha, 410000, China

Keywords: Intelligent Intervention Model, Ankle Injury Rehabilitation, Wearable Sensors.

Abstract: There are some problems in the traditional rehabilitation methods of ankle injury such as lack of individuation.

In this paper, intelligent intervention model is proposed, integrates multi-modal data, and constructs a full-

cycle closed-loop management system of "acquisition - analysis - decision - feedback".The data acquisition

layer uses wearable sensors and imag-ing technology to obtain ankle kinematic parameters and soft tissue

status; the intelligent analysis layer realizes accurate assessment with the help of deep learning algorithms;

the dynamic decision-making layer formulates personalized treatment plans according to patients'

characteristics; and the feedback implementation layer realizes accurate rehabilitation training with the help

of flexible robots, biofeedback systems, and intelligent protective gears. The model shows significant

advantages in ankle injury rehabilitation and prevention but still faces challenges such as data fusion.In the

future, intelligent intervention models are ex-pected to provide better medical services for patients.

1 INTRODUCTION

The ankle joint, as a key weight-bearing joint of the

human lower limb, has a complex structure and

various functions, which not only needs to maintain

the stability of standing and walking and adapt to the

needs of high-intensity sports such as jumping and

steering. However, it is this functional specialization

that makes it a high incidence of sports injuries. The

ankle joint consists of the tibia, the distal fibula, and

the talus, and is surrounded by ligaments, tendons,

and the joint capsule. Among them, the lateral

collateral ligaments (anterior talofibular ligament,

posterior talofibular ligament, and calcaneofibular

ligament) are weaker than the medial ligaments

because of their anatomical position, and they are

more prone to tearing or rupture due to excessive

inversion or eversion during sports. According to

statistics, ankle injuries account for 15%-30% of all

sports injuries, of which about 80% are lateral

ligament injuries, especially in basketball, soccer, and

other sports that require sharp stops and jumps are the

most common (RobsoN,H. E.,1988). According to

some studies, about 710,000 people suffer from ankle

sprains of different degrees every day. According to

the statistics of the Dutch Health Security

Administration, of the 120,000 ankle injuries

registered annually, 36% (43,000) require surgical

treatment at a later stage, with an average annual

treatment cost of nearly 40 million dollars. Minor

ankle sprains can be recovered with simple treatment,

while 70% of patients with acute sprains develop

ankle instability and recurrent sprains, and 20% to

40% of acute ankle sprains develop chronic ankle

instability (CAI) due to inappropriate treatment or

insufficient attention (Coronado, R. A.,2011).

Traditional rehabilitation methods are mainly based

on static assessment and doctor's experience to

formulate treatment plans, which have the problems

of insufficient personalization and untimely

monitoring of patient's dynamic changes, resulting in

unsatisfactory rehabilitation effect and a high

recurrence rate of injury. The intelligent intervention

model integrates multimodal data, covering

biomechanics, imaging, exercise physiology, and

other fields, which can realize accurate assessment

and personalized intervention for ankle injuries and

promote the change of ankle injury management

mode to the direction of precision and dynamization

(Zhang, Y.,2013).

This paper takes the "intelligent intervention

model" as the core research framework, and through

the deep integration of deep learning algorithms,

wearable biomechanical sensing technology and

digital biomimicry model constructs a full-cycle

Liao, Z.

Research Progress of Intelligent Intervention Modeling for Ankle Injury Rehabilitation and Prevention.

DOI: 10.5220/0014493900004933

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Biomedical Engineering and Food Science (BEFS 2025), pages 391-397

ISBN: 978-989-758-789-4

391

closed-loop management system covering "data

acquisition-intelligent analysis-dynamic decision-

making-real-time feedback". This system is designed

to solve the problem of traditional ankle joint

management. This system aims to solve the three core

problems of traditional ankle injury management:

lack of personalized rehabilitation programs (e.g.,

only 32% of patients receive biomechanically

adapted training programs), lagging in prevention

(less than 60% of high-risk movements identified

accurately), and sloppy prognosis assessment (static

imaging misdiagnosis rate as high as 18%).

2 CORE TECHNICAL

FRAMEWORK OF

INTELLIGENT

INTERVENTION MODEL

The intelligent intervention model is a kind of model

that uses artificial intelligence technology to realize

automatic intervention and optimization of specific

objects or systems through the analysis and

processing of data. Among them, for the medical

field, it can firstly assist doctors in formulating

personalized treatment plans by analyzing massive

medical data and achieving precise interventions in

chronic disease management, disease risk prediction

and other aspects.

The deeper level is its core technology

framework: the intelligent intervention model follows

the closed-loop operation logic of "acquisition-

analysis-decision-making-feedback", and its

technical architecture consists of multiple

interoperable modules:

2.1 Data Acquisition Layer

Wearable sensors, such as accelerometers,

gyroscopes, and pressure sensors, are utilized to

acquire the 3D kinematic parameters of the ankle

joint in real-time during movement, including

information such as angle, velocity, and acceleration.

Meanwhile, the combination of imaging techniques

such as weight-bearing CT and MRI can accurately

capture the state of the soft tissues of the ankle joint,

such as the damage to ligaments, tendons, and

cartilage. In this regard, this subsection focuses on

flexible wearable sensors. The acquisition, sensing,

and analysis methods of their motion signals include

electrophysiological signal-based monitoring (EMG

signals, ECG signals, EEG signals), photoelectric

sensing-based sign monitoring (heart rate, heart rate

variability, oxygen saturation), and electrochemical

biosensing-based monitoring (lactate, glucose).

Among the electrophysiological signals

monitored: EMG: Electrophysiological signals

generated by the nerve cells of the muscular system

are monitored through electrodes attached to the body

surface, which can be used for movement analysis

and early warning of muscle fatigue. Jianyong

Ouyang of the National University of Singapore has

achieved a good fit to the skin under wet skin

conditions by introducing an organic dry electrode

film based on sorbitol-modified poly(3,4-

ethylenedioxythiophene)-poly(styrenesulfonate)

composite with an electrical conductivity of 545 S-

cm-1. Electrocardiographic signals:

electrophysiological signals from the heart's regular

changes that are relatively easy to monitor. To solve

the problems of allergy and poor contact with

traditional electrodes, Chen et al. prepared

biocompatible composite electrodes by interfacial

polymerization of silk protein and polypyrrole. EEG

signals: Electrophysiological signals are generated by

the electrical activity of neurons in the brain, which

are mainly collected by multi-site point electrodes on

the surface of the scalp. Vörös et al. from ETH

Zurich, Switzerland, proposed a conductive-based

soft micropillar polymer electrode to realize high-

quality acquisition of EEG signals.

Photoelectric sensing monitoring: The core of

photoelectric sensing is a photoelectric sensor, which

is mainly used to monitor the changes in blood

volume by photoelectric volumetric tracing (PPG) to

realize the monitoring of human vital signs. Based on

organic phototransistors and inorganic LED doping,

Zhao Niobium of the Chinese University of Hong

Kong has developed ultra-thin flexible sensors, which

are applied to heart rate, pulse, blood pressure, and

other physiological signals detection.

Electrochemical biosensing monitoring:

Electrochemical sensors consist of a receptor and an

electrochemical conversion element, and are used to

detect metabolites, electrolytes, ions, and other

components in human body fluids. Jia et al. at the

University of California were the first to propose the

use of a tattoo electrode electrochemical sensor based

on flexible printing for the detection of sweat lactate

with a sensitivity of 220 nA-mM-1 (Su, B. T.,2022).

2.2 Intelligent Analysis Layer

Deep learning algorithm, Deep learning is a branch of

machine learning, since Geoffrey Hinton proposed

BEFS 2025 - International Conference on Biomedical Engineering and Food Science

392

the idea of deep learning in 2006, it has made

breakthroughs in the field of speech recognition,

image recognition, etc. CNNs are an important model

in deep learning, which is inspired by the hierarchical

processing mechanism of the biological visual cortex,

and it has a powerful feature learning and feature

expression capability.

The basic structure of CNNs includes a

convolutional layer, a pooling layer, and a fully

connected layer. The convolutional layer extracts

image features through convolutional operations, the

pooling layer reduces the feature dimensionality and

enhances the translational invariance of features

through aggregation statistical operations, and the

fully connected layer is used for classification or

regression tasks (Lu, H. T.,2016).

2.3 Dynamic Decision-Making Layer

Combining clinical practice guidelines and individual

patient characteristics, such as age, gender, level of

exercise, and injury history, the intelligent

intervention model can generate a personalized

treatment plan. Children with immature skeletal

development and epiphyseal injuries (e.g., Salter-

Harris typing) need to prioritize growth plate

protection (Xing, J. H.,2021). Smart technology

simulates skeletal stress distribution through 3D

biomechanical modeling to guide personalized brace

design and avoid joint stiffness caused by traditional

cast immobilization (Fan, J. P.,2017). Osteoporosis

and decreased healing capacity are core challenges

for elderly patients. Smart technology predicts the

risk of delayed postoperative bone healing and guides

the selection of anti-osteoporosis drugs through

digital bone density assessment combined with

genomic analysis (e.g., COL1A1 gene

polymorphism) (Cao, F.,2010).

2.4 Feedback Execution Layer

The feedback execution layer of the intelligent

intervention model is the final execution link in the

closed loop of "perception-analysis-decision-making-

feedback", and its core function is to transform the

decision-making scheme output from the intelligent

analysis layer into operable physical or digital

interventions and dynamically optimize the

intervention strategy through real-time data feedback.

In ankle injury rehabilitation, the feedback execution

layer realizes the dynamic adjustment and precise

execution of personalized rehabilitation training

through the integration of flexible robots,

biofeedback systems intelligent protective gears, and

other technologies. The core technology consists of

three points: flexible robot dynamic control: flexible

ankle rehabilitation robot (FARR) through the bionic

ligament design and multi-degree-of-freedom motion

control technology, real-time reception of the

patient's joint mobility (ROM) and

electromyographic signal (EMG) data, dynamic

adjustment of the joint torque and movement

trajectory (Zhang, W. Y., 2014; Yang, Z.,2005). The

real-time interaction of biofeedback system: virtual

reality (VR) augmented training system collects ankle

inversion angle and ground reaction force data

through wearable sensors (e.g., inertial measurement

unit IMU), and combines with CNN algorithm to

recognize abnormal gait patterns and generate

visual/auditory cues in real-time (Fan, Z. M.,2023).

Intelligent brace dynamic adaptation: intelligent

brace based on plantar pressure distribution and

sports scenarios, dynamically adjusting the hardness

of lateral ankle joint support through an air pressure

adjustment module (Yan, B.,2006).

3 PROGRESS IN THE

APPLICATION OF SMART

INTERVENTIONS IN

REHABILITATION

3.1 Application Progress of Wearable

Sensors for Ankle Injury

Rehabilitation

The progress of the application of wearable sensors

for ankle injury rehabilitation can be divided into

three main points: the type of movement signals, the

improvement of sensor performance, and scientific

exercise training. The types of sports signals can be

divided into sports biochemical signals, sports

electrophysiological signals, sports posture signals,

and bio-tissue dynamics signals, which contain

specific physiological sign information, and through

the real-time monitoring of these signals, it can be

realized to objectively evaluate the training effect and

the physical condition of the athletes.

As for the improvement of sensor performance, it

is due to the cross-fertilization of various disciplinary

fields that the performance and functions of sensors

have been significantly improved. For example,

sensors based on electrophysiological signals can

collect higher quality electrophysiological signals for

various types of sports analysis; sensors based on

Research Progress of Intelligent Intervention Modeling for Ankle Injury Rehabilitation and Prevention

393

photoelectric volumetric tracing methods can realize

the monitoring of multiple physiological indexes;

non-invasive detection based on electrochemical

sensors has made great progress.

Finally, scientific exercise training requires a

complete exercise monitoring system, which requires

the integration of multiple flexible wearable exercise

sensors. It will be possible to build a multifunctional

sports monitoring platform, which will have the

functions of monitoring physiological indexes of

sports training, analyzing sports technology and

tactics, analyzing sports psychological conditions,

and predicting sports injuries (Su, B. T.,2022).

3.2 Progress of Deep Learning

Algorithms for Ankle Injury

Rehabilitation

Convolutional Neural Networks (CNN), as one of the

core technologies of deep learning, have shown

significant advantages in the fields of medical image

analysis, sports biomechanics modeling, and

intelligent rehabilitation equipment in recent years.

Its application in ankle injury rehabilitation mainly

focuses on image diagnosis, dynamic monitoring,

personalized rehabilitation program development,

and intelligent equipment control.

CNN-based smart insoles and inertial sensors

(IMUs) can collect ankle kinematic parameters (e.g.,

inversion angle, ground reaction force) in real time

and realize dynamic risk assessment through feature

extraction. Studies have shown that CNN models

fused with multidimensional biomedical data (heart

rate, EMG signals) can recognize high-risk

movements such as sharp stops and jumps with 85%

accuracy. For example, to address the risk of ankle

inversion in basketball players, the system can analyze

the plantar pressure distribution through CNN to

trigger real-time warnings and guide movements.

The flexible ankle rehabilitation robot combines

CNN algorithms to achieve precise control of multi-

degree-of-freedom movements. Through bionic

ligament design and impedance adjustment, its joint

torque fluctuation is 40% lower than that of

traditional equipment, and its workspace covers 98%

of the physiological activity range. The brain-

computer interface foot and ankle rehabilitation robot

from Tsinghua Changgeng Hospital further integrates

EMG signals and eye tracking, utilizes CNN for

multimodal data fusion, and significantly improves

the patient's active mobility and balance function of

the ankle joint, with a gait abnormality improvement

rate of 73% at 20 days post-surgery (Lu, H. T.,2016).

3.3 Personalized Rehabilitation

Training

For complex ankle fractures (e.g., Weber type C), 3D-

printed guides assist in the precise placement of

internal fixation screws, reducing intraoperative

fluoroscopy time and soft tissue stripping (Liu, S.

H.,2018). In postoperative rehabilitation, the flexible

ankle robot balances joint stability and bone healing

needs through low-intensity progressive training

(torque fluctuations reduced by 40%) (Wang, X.

Z.,2016).

Severe injuries (e.g., large talar cartilage defects):

arthroscopic bone graft repair is used to implant

autologous cancellous bone to promote cartilage

regeneration, and postoperatively, with weight-

bearing CT to assess the effect of force line

correction, avoiding the trauma of traditional

osteotomy (Zhang, Z. H.,2012).

Intelligent protective gear recommendation

system for amateur athletes based on arch type and

sports scenarios (e.g., running, basketball) reduces

the probability of inversion injuries by 37%

Professional athletes combine dynamic load

monitoring with platelet-rich plasma (PRP)

injections, which promotes cartilage repair while

adjusting the training intensity to shorten the time to

return to the field of play by 30%( Su, D. Y.,2025;

Ming, P. J.,2017).

3.4 Specific Application of the Three

Core Technologies of the Feedback

Execution Layer

Flexible ankle rehabilitation robot for patients with

ankle dorsiflexion limitation, the robot can adaptively

reduce the resistance in the direction of plantarflexion

based on the kinematic model, and at the same time

increase the dorsiflexion auxiliary torque, so that the

joint torque fluctuation is reduced by 40% compared

with the traditional equipment (Zhang, W. Y., 2014;

Yang, Z.,2005). Virtual reality (VR) augmented

training system: when the patient's inversion angle

exceeds the safety threshold (>15°), the system

triggers a red alert and suspends training until the

patient adjusts to the correct posture (Fan, Z.

M.,2023). Intelligent protective gear based on plantar

pressure distribution and sports scenarios: when a

basketball player stops sharply and jumps, the

protective gear automatically enhances the hardness

of lateral support (up to 80 kPa), reducing the

probability of inversion injury by 37% (Yan,

B.,2006).

BEFS 2025 - International Conference on Biomedical Engineering and Food Science

394

4 TECHNICAL CHALLENGES

AND FUTURE DIRECTIONS

4.1 Existing Technical Challenges and

Future Research Directions for

Lower Limb Exoskeleton Robots

Although great progress has been made in the field of

intelligently powered lower limb exoskeleton robots,

they still face many challenges: firstly, the joint

actuators of intelligently powered lower limb

prostheses have problems such as large weight and

small output torque, which limit the motion

characteristics of the joints of the prostheses. Then

there is insufficient research on the correlation control

of multi-joint prostheses, which affects the stability,

safety, and comfort of disabled people in different

walking gaits. Secondly, although the accuracy rate

of motion intent recognition is high, there is still a risk

of falling in practical applications, and there are

limitations in the existing sensing methods. At the

same time, there is insufficient research on motion

intention recognition in complex dynamic

environments, and the influence of environmental

changes on the recognition model is significant.

Finally, the stimulation device of the sensory

feedback system is simple, the variety of stimulation

modes is limited, and the sensing alternatives are not

natural enough and need long time training.

In the future, the bionic structure design of

intelligent powered prostheses and human-machine

integration for complex environments will become

the focus of research. Clinical trial-driven control

algorithms, sensing methods, and perceptual

feedback will become new research highlights in this

field. Nevertheless, the research on smart-powered

lower limb prostheses still needs to solve many basic

scientific problems and technical difficulties to

realize more natural and smooth motion control and

better human-machine integration (Wang, Q.

N.,2016).

4.2 Current Shortcomings of Deep

Learning Algorithms and Future

Optimization

Firstly, regarding data standardization and privacy

protection, the integration of cross-modal data

(imaging, genome, exercise physiology) requires

unified protocols (e.g., DICOM, HL7), and although

blockchain technology can encrypt the storage, the

real-time processing efficiency still needs to be

improved (Meng, C. B.,2004; Mu, S., Cui,2021;

Yuan, Y.,2016).

Secondly, algorithm generalization ability is

improved, existing CNN models are limited in

performance in small sample data (e.g., rare injury

types), and need to be combined with migration

learning and generative adversarial networks (GAN)

to enhance generalization (Wang, K. F.,2017). Finally

regarding clinical translational validation, most

studies are limited to single-center trials, and

multicenter randomized controlled trials (RCTs) are

needed to validate long-term efficacy (Liu, G.

B.,2017). For example, the cumulative effect of

carbon plate running shoes on ankle joint loading in

non-professional athletes needs to be supported by

more than 10 years of cohort studies.

Its future directions mainly include: the fusion of

regenerative medicine and CNN: 3D-printed bionic

cartilage scaffolds combined with PRP injections,

optimizing the scaffold porosity and matching

mechanical properties through CNN (Sun, X.,2006;

Wang, L.,2014); group intelligence platform:

building a collaborative network in the cloud,

integrating data from doctors, patients and

rehabilitators, and realizing the optimization of

dynamic solutions (He, Y.,2019)s.

4.3 Deficiencies and Optimization of

Personalized Rehabilitation

Programs

Firstly, for data standardization: cross-modal data

(imaging, genomic, exercise physiology) integration

needs to be a unified protocol to avoid algorithmic

bias (Meng, C. B.,2004; Mu, S., Cui,2021). Secondly,

in the human-robot interaction bottleneck, the

existing rehabilitation robots are not flexible enough,

and so on, bionic materials (e.g., shape memory

alloys) need to be developed to improve comfort.

Finally, regarding ethics and privacy, genetic data and

motor biometrics need to be encrypted and stored in

compliance with the Code of Practice for Medical

Data Security Management.

4.4 Technical Challenges and

Innovative Directions for the

Feedback Execution Layer

First, multimodal data fusion bottleneck: the existing

system has insufficient cross-modal fusion capability

for imaging (e.g., MRI), exercise physiology (EMG),

and genomic data, and a heterogeneous data mapping

model based on graphical neural networks needs to be

Research Progress of Intelligent Intervention Modeling for Ankle Injury Rehabilitation and Prevention

395

developed to improve the biological suitability of

intervention strategies. Secondly, the optimization of

human-robot interaction flexibility: the joint impact

of traditional rigid robots is prone to secondary

damage, and variable stiffness actuators based on

magnetorheological fluid need to be explored in the

future to achieve smooth torque transition

(fluctuation <5 Nm). Finally, a long-term efficacy

validation system is missing: most studies rely on

short-term laboratory data (<6 months), and a

multicenter follow-up platform (e.g., blockchain-

based healthcare data consortium) needs to be

established to validate the 10-year cumulative risk

impact of smart interventions on traumatic arthritis

(Zhang, W. Y., 2014; Yang, Z.,2005; Fan, Z.

M.,2023; Yan, B.,2006).

5 CONCLUSION

The intelligent intervention model brings new

opportunities for ankle injury rehabilitation and

prevention by integrating multidisciplinary

technologies. It demonstrates significant advantages

in accurate diagnosis, personalized rehabilitation

training, and injury prevention, effectively improving

the accuracy and efficiency of ankle injury

management. However, technical challenges such as

data fusion, human-computer interaction, and clinical

validation still need to be overcome to realize the

widespread clinical application of intelligent

intervention models. In the future, with the

continuous innovation of technology and in-depth

interdisciplinary intersection, the intelligent

intervention model will develop in the direction of

more personalized, intelligent, and universal,

providing better medical services for patients with

ankle injuries.

REFERENCES

Cao, F., Shen, B., Li, Y., Huang, Q., Yang, J., Zhou, Z. K.,

Kang, P. D., Peng, W. Z., Xia, Q. J., Pei, F. X. (2010).

Sichuan da xue xue bao. Yi xue ban = Journal of

Sichuan University. Medical science edition, 41(5),

831–835.

Coronado, R. A., Wurtzel, W. A., Simon, C. B., Riddle, D.

L., George, S. Z. (2011). Content and bibliometric

analysis of articles published in the Journal of

Orthopaedic & Sports Physical Therapy. The Journal of

orthopaedic and sports physical therapy, 41(12), 920–

931.

Fan, J. P., Liu, X. W., Zhao, T. Y., et al. (2024). Research

progress on applications of artificial intelligence in

orthopedic surgical practice. Journal of Spinal Surgery,

22(2), 135 140.

Fan, Z. M., Wang, C. B., Xie, T., et al. (2023). Effects of

limb coordination assistive devices combined with VR

training on gait balance and surface electromyography

in patients undergoing intracranial vascular

interventional therapy for acute cerebral infarction.

Chinese Journal of Stroke, 18(2), 194 200.

He, Y., Li, H. X., Yang, Y. (2019). Dynamic mechanisms

and guidance strategies for online group polarization in

new media environments: A case study of content

intelligent distribution platforms. Information Science,

37(3), 146 151+168.

Liu, G. B., Zhang, G. P., Ren, Q. Y., et al. (2017).

Diagnostic value of MRI for ligament and tendon

injuries around the ankle in different postures: A single

center diagnostic trial. Chinese Journal of Tissue

Engineering Research, 21(4), 598 602.

Liu, S. H., Zhang, B., Liu, Q., et al. (2018). Application of

3D printing technology combined with occlusal plane

positioning mandibular angle osteotomy guide

protector in mandibular angleplasty. Chinese Journal of

Aesthetic and Plastic Surgery, 29(9), 527 531.

Lu, H. T., Zhang, Q. C. (2016). A review on applications of

deep convolutional neural networks in computer vision.

Journal of Data Acquisition and Processing, 31(1), 117.

Meng, C. B., Zhang, J. W. (2004). Analysis and

implementation of data exchange between HL7 and

DICOM. Medical Information, (12), 787 793.

Ming, P. J., Li, J., Liu, T., et al. (2017). Experimental study

on a novel ankle protective brace for preventing landing

injuries in paratroopers. Journal of Preventive Medicine

of Chinese People's Liberation Army, 35(11), 1361

1370.

Mu, S., Cui, M., Huang, X. D. (2021). Data integration

methods for multimodal learning analytics in

panoramic perspectives. Modern Distance Education

Research, 33(1), 26 37+48.

RobsoN,H. E. (1988). British Journal of Sports Medicine--

historical development. British journal of sports

medicine, 22(4), 130–131.

Su, B. T., Li, J. L., Xu, H. H., Xu, Z., Meng, J. X., Chen, X.

P., Li, F. Y. (2022). Scientific training assistance:

Applications of flexible wearable sensors in sports

monitoring. Scientia Sinica Informationis, 52(1), 54 74.

Su, D. Y., Ma, Y. B., Liu, J. L., et al. (2025). Application

value of dual energy CT imaging in animal bone repair

evaluation. Chinese Journal of Comparative Medicine,

35(1), 155 162.

Sun, X., Xi, T. F. (2006). Advances in biomaterials and

regenerative medicine. Chinese Journal of Reparative

and Reconstructive Surgery, (2), 189 193.

Wang, K. F., Gou, C., Duan, Y. J., et al. (2017). Research

progress and prospects of generative adversarial

networks (GANs). Acta Automatica Sinica, 43(3), 321

332.

BEFS 2025 - International Conference on Biomedical Engineering and Food Science

396

Wang, L.,Dai, K. R. (2014). Individualized orthopedic

treatment and 3D printing technology. Journal of

Medical Biomechanics, 29(3), 193 199.

Wang, Q. N., Zheng, E. H., Chen, B. J., et al. (2016).

Current status and challenges of intelligent powered

lower limb prostheses for human machine integration.

Acta Automatica Sinica, 42(12), 1780 1793.

Wang, X. Z., Li, H. Y., Wang, Y. C., et al. (2016). High

precision adaptive backstepping control for flexible

joint robots. Information and Control, 45(1), 17.

Xing, J. H., Fan, T. S., Li, Z. J., et al. (2021). Efficacy

analysis of closed reduction and splint fixation for

Salter Harris type II distal radial epiphyseal injuries in

children. Journal of Tianjin University of Traditional

Chinese Medicine, 40(5), 600 603.

Yan, B. (2006). Working principles and motion simulation

of automatic adjustment mechanisms for pneumatic

disc brake clearance.Master's thesis, Wuhan University

of Technology.

Yang, Z. (2005). Research on robot compliance control

based on impedance control.Master's thesis, Southeast

University.

Yuan, Y., Wang, F. Y. (2016). Blockchain technology:

Current status and future prospects. Acta Automatica

Sinica, 42(4), 481 494.

Zhang, W. Y., He, J. K., Liu, Y. X., et al. (2014). Design

and fabrication of ligament bone scaffold bionic

interface. Journal of Xi'an Jiaotong University, 48(10),

137-142.

Zhang, Y., Poon, C. C., MacPherson, E. (2013). Editorial:

an inaugural message for the IEEE Journal of

Biomedical and Health Informatics. IEEE journal of

biomedical and health informatics, 17(1), 3.

Zhang, Z. H., Liu, Z. L., Gao, Z. Z., et al. (2012). Selection

and application of bone repair substitute materials for

bone defect reconstruction. Chinese Journal of Tissue

Engineering Research, 16(52), 9836 9843.

Research Progress of Intelligent Intervention Modeling for Ankle Injury Rehabilitation and Prevention

397