Upper-limb amputations significantly affect independence and quality of life, particularly in low-income regions where advanced prosthetic technology is costly and lacks adequate personalization. Conventional myoelectric prostheses, while offering functional restoration, have limited adaptability and high cost. This study presents a personalized transradial myoelectric prosthesis that combines additive manufacturing and Artificial Intelligence (AI) control, offering an accessible and high-performance solution. The prosthesis design utilizes additive manufacturing (3D printing) for anatomical personalization via 3D scanning and parametric modeling. An AI-driven control system utilizes machine learning to classify electromyography (EMG) signals in real-time, specifically detecting the user’s intention to perform flexion or extension movements, and tailoring responses to individual users. Evaluation employed the “Brief Activity Measure for Upper Limb Amputees (BAM-ULA)” protocol with nine participants with transradial amputations. Trials with the nine participants yielded an average BAM-ULA score of 7.4 out of 10 (Standard Deviation (SD) 0.7). This demonstrated robust functional performance, comparable to high-end commercial devices in initial tests. Gross motor tasks saw 100% success rates; fine motor tasks, 22.2%. Integrating AI and additive manufacturing resulted in an affordable, high-performance, personalized prosthesis. This work highlights how localized digital manufacturing enables accessible customization for users in low-resource settings. The main novelty is this validated integration of personalized additive manufacturing and adaptive AI control in an affordable transradial prosthesis addressing the needs of developing countries.
Autor(es):Romero, Enzo
Garcia, Jose G.
Parra, Magno
Caballa, Sebastian
Saldarriaga, Alejandro M.
Luque, Edson F.
Rodriguez, Dante J.
Abarca, Victoria E.
Elias, Dante A.
Año: 2025
Título de la revista: IEEE Acces
Ciudad: Lima, Perú
Volumen: 13
Página inicial - Página final: 139631 - 139649
Url: https://doi.org/10.1109/ACCESS.2025.3596475