Machine Learning for Pediatric Mobility Therapies
Abstract
An estimated 4.3% of children in the United States have disabilities, a prevalence that has been steadily rising in recent years. Children with motor disabilities face significant delays in their ability to walk independently; many do not achieve this milestone until they are between 3 to 5 years old, which can be 2 to 4 years later than their nondisabled peers. Moreover, many do not receive access to wheeled mobility devices until well into their preschool years, significantly hindering their opportunities for independent movement and social interaction. Early access to powered mobility devices is essential, as it enables children to explore their environment and engage in activities that foster social cognitive development.
Effective assessment and intervention strategies are critical for children with motor disabilities, as early diagnosis and personalized therapeutic plans can significantly improve developmental outcomes. The Assessment for Learning Powered Mobility (ALP), used in pediatric hospitals, provides valuable insights into motor, cognitive, and social-emotional growth by evaluating how children interact with assistive devices. However, there is a growing need to refine diagnostic tools by leveraging data-driven approaches. In this study, we utilize the Permobil Explorer Mini, a powered wheelchair designed specifically for young children. We gather data from nine participants with motor disabilities. These findings highlight the potential for predictive models to enhance diagnostic accuracy and improve therapeutic interventions for children with disabilities. Ultimately, this data-driven approach will enable clinicians to anticipate and address the therapeutic needs of children with disabilities more accurately, ensuring timely and effective interventions that promote mobility and independence.
