Transfers are an essential task performed numerous times a day by wheelchair users with lower limb paralysis or dysfunction. Because transfers are highly physically demanding on the upper limbs they are a major contributor to the development of upper limb pain and injury. In previous work, we developed an instrumented transfer station to evaluate the biomechanics of sitting-pivot wheelchair transfer techniques. We discovered that using certain prescribed trunk and hand positions reduces loading on the upper limb joints. We also developed and established the psychometric properties of the Transfer Assessment Instrument (TAI) which is a quick and easy way that clinicians can determine how well their patients adhere to best transfer practices. We found that subjects who had TAI detectable technique deficits and received individualized training by a therapist to correct these deficits showed an immediate improvement in their movement patterns and forces after the training session. While one-on-one training with a therapist is ideal, the training is labor intensive. Moreover, shorter rehab stays and heavy caseloads prohibit clinicians from spending a lot of time teaching and reinforcing transfer skills. Once taught, patients are generally responsible for practicing and implementing the techniques on their own. Supplementing the clinician's training with an interactive coaching tool that provides veterans with a way to check and correct their form on their own would likely facilitate the learning and retention of new transfer skills. This pre-development phase of research will determine the types of transfer movement variables that can be reliability detected with the Microsoft KinectTM, a low-cost, markerless motion capture system and the accuracy associated with each movement variable.
Aim 1 will be to identify Kinect motions that are reliable and consistent with motions measured by a reference 3D marker based system (e.g. Vicon). Twenty-five wheelchair users who are independent with transfers will transfer to/from two surfaces (e.g. a level bench and commode) while the two systems simultaneously record their movement patterns. We hypothesize that the 3D trunk, shoulder, elbow and wrist angle motions measured with the Kinect will follow a similar trajectory as the Vicon-measured motions and the peak angles will be repeatable and similar between the two systems.
Aim 2 will evaluate an additional 25 subjects to explore the relationships between TAI scores and Kinect motions during transfers. We hypothesize that Kinect motions will be significantly correlated with the quality of the transfers. Successful completion of this study will allow for defining the design criteria and features of the transfer coaching system.