Toward Intelligent Adaptive Transfemoral Prostheses: Bridging Machine Learning, Biomechanics, and Clinical Mobility

9:15 AM - 10:30 AM Tue, 23 Mar

Description

Symposium topic

Adaptive control remains one of the greatest challenges in transfemoral prosthetics. While microprocessor-controlled knees have improved mobility and safety, current systems have limited capacity to recognise changing environments and predict user movement in real time. Emerging advances in machine learning, biomechanical modelling, and wearable sensing technologies provide opportunities to develop intelligent prosthetic systems capable of identifying locomotor intent, recognising walking surfaces, and adapting knee behaviour accordingly. This symposium explores recent developments in hybrid machine learning approaches for surface recognition and knee kinematics prediction (Figure 1) and discusses their implications for future adaptive prosthetic systems and clinical practice.

Underpinning science and theory

The symposium is grounded in evidence from biomechanics, rehabilitation engineering, machine learning, electromyography, and musculoskeletal modelling. Presentations will draw upon recent research demonstrating the application of hybrid neural network architectures, including MLP-SVM and deep learning models, for locomotion intent detection, surface classification, and continuous knee kinematics prediction. Biomechanical validation using musculoskeletal modelling will be highlighted to ensure physiological plausibility and clinical relevance.

Relevance for the ISPO World Congress

This symposium aligns with the ISPO 2027 theme, “A World Connected: Advancing Assistive Technologies for Global Equity.” Intelligent prosthetic technologies have the potential to improve mobility, safety, and participation for people with transfemoral limb loss worldwide. The symposium will highlight pathways for translating advanced assistive technologies from research laboratories into clinical practice and education.

Perspectives

The symposium will present four complementary perspectives: (1) machine learning approaches for locomotion intent detection and surface recognition; (2) deep learning methods for predicting knee joint kinematics and validating outcomes through musculoskeletal modelling; (3) clinical implications for prosthetic rehabilitation and mobility outcomes; and (4) educational and workforce considerations for integrating artificial intelligence into future prosthetic and orthotic practice.

Learning objectives

Participants will be able to: (1) describe current challenges in adaptive transfemoral prosthesis control; (2) explain the role of machine learning in locomotion intent detection and surface recognition; (3) discuss approaches for predicting knee kinematics using deep learning and musculoskeletal modelling; (4) evaluate opportunities and limitations for clinical implementation of intelligent prosthetic systems; and (5) identify future educational, research, and clinical priorities for advancing adaptive prosthetic technologies globally.

Presenter

Information about the program