Implementation of a sensor platform for multi-day measurement of manual wheelchair user mobility patterns in real-world environments to inform clinical training and improved contexts for research - In the United States, there are more than 3.6 million individuals who rely on the manual wheelchair (MW) for almost all academic, vocational, and societal activities. Despite this substantial stakeholder population, modern decision-making around MW training interventions and MW prescription is largely a clinical art that leans heavily on self-reported data and trial-and-error. Furthermore, the body of MW research, like many disciplines of clinical science, struggles with extending controlled laboratory findings to real-world representation. The key element missing is a clinician-accessible means of collecting objective data on the mobility patterns of each unique manual wheelchair user (MWU), as the real-world user needs and habits best dictate the training interventions and equipment that should be prescribed. For the Switzer Research Fellowship, I propose to validate and implement a sensor platform that serves as a clinical assessment and research tool by characterizing multi-day, real-world mobility patterns of MWUs, which coincides with NIDILRR’s domain outcome of “health and function”. This proposed research builds on my prior doctoral work and will accomplish one of three essential research components needed to fulfill my long-term goal: informing the selection of MWs that maximize individual MWUs’ functional mobility, thus improving their health and independence. To accomplish my proposed research, I will leverage existing IMU sensor technology and wheelchair monitoring methodologies from a collaborator to optimize a combined sensor platform based on performance metrics critical to multi-day measurement of wheelchair kinematics and surface slope (e.g. sensor placement, accuracy, drift). This sensor platform will be reviewed by two focus groups of 5 clinicians to identify desirable data outputs, and concurrently validated against camera recordings of drive wheel motion during Wheelchair Skills Test (WST) propulsion tasks performed by 12 MWUs across a wheelchair skills course. Following sensor platform validation, I will collect 12 MWUs’ wheelchair kinematic data across 7-10 days in their real-world environments, apply machine learning clustering to identify the predominant bouts of movement, and compare the maneuvers that comprise the identified bouts with those described in the WST. This approach will inform not only the frequency of different bout types in the real-world, but also how well the maneuvers conducted in the WST agree with those used in real-world bouts. Successful completion of this proposal will yield initial, clinically-useful insights into how MWUs navigate their environments, as well as a potential clinical research/assessment tool for customizing MWU training and informing improved ecological validity in the design of future MW research studies.
