Evaluating the Feasibility of Force-Based and Position-Based Prosthetic Control for Children Born without a Hand

/Abstract Congenital upper limb deficiencies are the most common reason for limb absence in children and occur in approximately 1 in 500 live births. Although many children are prescribed a prosthetic arm to offset functional deficits, abandonment rates among pediatric users range as high as 45%, nearly double the rate of adult abandonment. A main driving factor for this is that highly functional adult prosthetic control systems—namely pattern recognition and machine learning methods—have not yet been translated to children due to a belief that they may not have the control over their residual muscles required to use such systems. Our research says otherwise, finding that 9 children born without a hand were successfully able to imagine moving their affected muscles into 10 unique grasps, and resulting in highly classifiable data. Therefore, my long-term research goals are to (1) define the innate capabilities of children born with upper limb deficiencies and (2) leverage those innate abilities to help develop more dexterous pediatric prosthetic devices. I will work towards these goals by fusing two existing prosthetic control systems: the long-established surface electromyography (sEMG), which measures the electrical activity of the muscles, and the recent advancement of sonomyography (ultrasound, US), which images the muscle deformation within the arm. Individually, our work has shown that these systems, in conjunction with machine learning algorithms, can predict missing hand grasp intent with accuracies of 70-99%, but they have never been used together. It remains unknown the degree to which their complementary information can be used to exceed what each system can achieve individually, and thus offer more advanced control techniques such as proportional force or proportional position control. I will pursue the following specific aims: (1) Maximize the combined static state prediction accuracy of sEMG and US using a naïve Bayesian fusion approach, and (2) Investigate the feasibility of pediatric proportional control by building regression models to predict a spectrum of desired grasp force and position values. My central hypothesis is that children will have a considerable level of proportional control over their residual muscles, reflected in high classification accuracies and low regression errors, and that fusing sEMG and US will result in significantly higher classification accuracies than either system can achieve in isolation, indicating a high level of complementary function. By the end of this proposal, we will have a much richer understanding of children’s innate control and the nuanced information encoded in their affected muscle activity. This will result in better classification of motor intent for prosthetic control and establish the feasibility of incorporating advanced, proportional control techniques into pediatric prosthetic limbs. We will also have built the first dataset that examines the ability of children to modulate both intended force and position, and that records time-synchronized data of sEMG and US working together. This proposal will serve as the foundation on which highly functional, pediatric prostheses can be designed. Project Number: 1F31HD118671-01 | Fiscal Year: 2025 | NIH Institute/Center: Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) | Principal Investigator: Eden Winslow | Institution: UNIVERSITY OF CALIFORNIA AT DAVIS, DAVIS, CA | Award Amount: $42,895 | Activity Code: F31 | Study Section: Special Emphasis Panel[ZRG1 F10B-Q (20)] View on NIH RePORTER: https://reporter.nih.gov/project-details/1F31HD11867101