D3 Research and Development




Overview of Research and Development Activities

State of the Science on Spinal Cory Injury

Research Projects: Effect of Local Cooling | Effects of Weight Shifting | Handrim Technology

Development Projects: Inflammation Modeling | Low Shear, Cool Cushion | Propulsion Training Tools


D3 Development of measurement tools for propulsion training in the natural environment

Task Leader(s): Dan Ding, Ph.D.

Co-Investigator: Michael Boninger, MD., Annmarie Kelleher, MS

Other participants: TBN GSR



Long-term use of the upper limbs for performing daily activities has led to an increase in the prevalence of musculoskeletal injuries and reports of pain. The Consortium for Spinal Cord Medicine recently published, Preservation of Upper Extremity Function Following Spinal Cord Injury: A Clinical Practice Guideline for Health Care Professionals. This monograph provides concise ergonomic and equipment recommendations based on review of relevant research.. The guidelines recommend reducing the frequency of repetitive upper limb tasks, minimizing forces required to complete tasks and minimizing extremes of wrist and shoulder motions. These guidelines also include specific recommendations related to wheelchair propulsion technique and transfer activities. The Clinical Practice Guideline is based on studies of objective measures of physical activities that were determined in laboratory and clinical settings.

The primary objective of this project is to develop and validate accelerometry-based field measures for wheelchair propulsion in the natural environment. We will examine the relationship between biomechanical variables and accelerometry measures during wheelchair propulsion in a laboratory setting. We will also assess the field utility of these measures to determine effectiveness of a wheelchair propulsion training program. The overall goal is to provide an effective tool to monitor actual upper extremity usage. This will enable us to more effectively evaluate propulsion interventions and understand of the etiology of upper extremity injuries and pain.


Specific Aims

This project will begin in the third year and will be utilize an unobtrusive wheelchair propulsion monitoring device (WPMD) capable of recognizing activities such as independent wheelchair propulsion, being pushed, driving, and other ADLs. The WPMD is being developed with support from the PVA Research Foundation for a two-year grant (Feb 2007-Jan 2009). WPMD will be able to recognize (and record) propulsion patterns including "semicircular," "looping over propulsion," and "arc." The WPMD will integrate three components. including a wearable 3-axis accelerometer (ACC) that monitors upper limb movements, a wheel rotation datalogger clipped onto the wheel and a vibration datalogger underneath the seat that monitors wheelchair motion. The proposed project is to further develop and validate ACC-based field measures that can capture wheelchair propulsion characteristics in terms of propulsion force, smoothness, cadence, and stroke length under free-living conditions based on the WPMD. The project will be implemented in two phases with the first phase focusing on the development of these measures and the second phase on validation of them. Figure 1 illustrates the various objectives of the PVA project, the proposed project, and future directions.

Subjects will propel a wheelchair on a computer-controlled dynamometer and along ADL courses, respectively, with a WPMD attached to the wheelchair and their upper limbs. Wheelchair propulsion biomechanics will be determined using the SmartWheel (Three Rivers Holdings, Mesa, AZ). We will evaluate a variety of data reduction strategies to determine how ACC-based measures can best represent wheelchair propulsion characteristics. The laboratory testing will also be followed by a 2-day field trial in the homes and community environments of the participants.

Aim 1: To validate the WPMD's performance in quantifying the amount of upper limb usage for wheelchair propulsion and further develop reliable ACC-based field measures for characterizing wheelchair propulsion in terms of propulsion force, smoothness, cadence, and stroke length.

Hypothesis 1a: The WPMD will be able to identify wheelchair propulsion episodes from a mix of ADLs with at least 90% accuracy during the laboratory trial.

Hypothesis 1b: ACC-based field measures collected by the WPMD will be able to predict greater than 60% of the variance of each wheelchair propulsion biomechanical variable (propulsion force, smoothness, cadence, stroke length).

Hypothesis 1c: The ACC-based field measures collected during the laboratory trial will be highly correlated (r>=0.8) with those collected during the home trial.


Aim 2: To examine the test-retest reliability of ACC-based field measures and the validity of the measures in assessing a wheelchair propulsion training program.

Hypothesis 2a: The test-retest reliability of ACC-based field measures indicated by the intraclass correlation coefficient (ICC) will be greater than 0.8 and there will be no significant difference between the test-retest data during Phase I and Phase II.

Hypothesis 2b: The ACC-based field measures collected during the laboratory trials before and after the training session will be significantly different if the training session is proved to be effective by the biomechanical variables.

Hypothesis 2c: There will be a significant change in the ACC-based field measures collected during the home trials in Phase I (before the training session) and Phase II (after the training session).


Aim 3: To quantify the amount of use and quality of upper limb movements in the home and community environment before and after the wheelchair propulsion training program.

Research Question #3a: How much time is usually spent on active wheelchair propulsion daily during the home trials before and after the training program?

Research Question #3b: How much time is usually spent on functional upper-limb use in addition to wheelchair propulsion?

Research Question #3c: Is the amount of use and quality of upper-limb movement during wheelchair propulsion related to the reported upper-limb pain?



We will recruit 26 manual wheelchair users with SCI to participate in this study. Inclusion criteria include: 1) 18 years of age or greater; 2) use of a manual wheelchair for primary mobility (> 80% of ambulation); 3) SCI at T2 level or lower. Exclusion criteria include: 1) inability to tolerate sitting for 2 hours; and 2) presence of upper limb pain that limits mobility. The study will be implemented in two phases. Each subject will complete the protocol described in the table below.



Protocol Overview

Phase I - Visit I

  1. Consent form, demographics/wheelchair survey, pain questionnaires
  2. WPMD attached to the wheelchair and the upper limb of the participant
  3. Laboratory trial of wheelchair propulsion on a computer-controlled dynamometer
  4. Laboratory trial along ADL courses
  5. 2-day field trial at home and community environment

Phase II - Visit II

  1. Laboratory trial along ADL courses
  2. Wheelchair propulsion training session
  3. Laboratory trial along ADL courses
  4. 2-day field trial at home and community environment


Anticipated Findings and Implications of Future Research

We anticipate that the ACC-based field measures will enable us to accurately and reliably capture wheelchair propulsion characteristics. We also anticipate that the WPMD will become an accepted clinical tool for recording the amount and quality of functional upper-limb movements. This information will be useful to clinicians and researchers to evaluate training outcomes and understand the etiology of upper limb injuries and pain. We expect to learn how manual wheelchair users are using their wheelchairs and upper limbs during daily activities of living, and whether they will be able to apply what they learned during the training program to their everyday wheelchair propulsion activities.

Our primary priorities for future research include: 1) a multi-center evaluation project that expands the subject pool, refines and validates the ACC-based measures for characterizing wheelchair propulsion among individuals with SCI, and examines the carry-over effects of different training programs under free-living conditions; 2) a larger-scale study of the relationship between the amount and quality of upper-limb movements, and the upper-limb pain and injury among this population. Additional future directions may include investigation on the effectiveness of sensor combinations (e.g., accelerometer and EMG) to capture more propulsion characteristics, and improve the current WPMD to collect data for longer durations, provide coaching functions and prevent secondary musculoskeletal injuries related to wheelchair propulsion.


Figure 1

Figure 1 - Relationship between D3 activities, current, and future work


Project Updates


We are recruiting research participants. You may be eligible to participate if you are a manual wheelchair user with a spinal cord injury and over the age of 18 years old. Find out more information in the recruitment flyer: .DOC | .PDF

We have tested 24 subjects so far. We have also been working on the data analysis. We have developed an algorithm to extract the number of strokes and calculate propulsion cadence over propulsion trials from the acceleration signals. The calculated stroke number and cadence from each accelerometer was compared with that from the criterion (Smart Wheel). Mean absolute error (MAE) and mean absolute percentage of error (MAPE) were calculated between the estimated and criterion values. Intraclass correlation coefficient were also calculated to assess the agreement. The preliminary results showed reasonable accuracy of the estimated temporal parameters using accelerometers, especially the one placed on the upper arm where the overall MAPE was 10.0% for stroke number and 10.7% for cadence, and ICC of 97% for stroke number and 82% for cadence. We have developed an activity classification model using a machine learning algorithm. The model showed an accuracy of 82.7% for propulsion, 80.4% for being pushed, 75% for activities of daily living (ADLs) and 75% for rest. Data collected from the natural environment is being evaluated using this model. We will continue working on data analysis for the effect of propulsion training, and estimation of other propulsion parameters such as peak forces and moments.


We have been working with the IRB office on the testing protocol until it was approved on March 1, 2011. Since then we started subject testing. So far 6 subjects completed the study protocol and we scheduled the 7th subject on June 1. We have also been working on the data analysis. We have developed an algorithm to extract the number of strokes over propulsion trials from the acceleration signals. The preliminary results showed the correlation between the estimated stroke number from the activity monitor and criterion stroke number observed in the video was highly correlated (r=0.99, p<0.001). We are currently working on estimating other propulsion parameters such as peak forces and moments using the acceleration signals from the activity monitors.


Over the past two months, we have been working on preparing and submitting the IRB documents. The documents are currently in review with the Pittsburgh VA Healthcare System and will be submitted to Pitt IRB upon the approval from the VA. We are also testing the instruments and setting up experimental protocol for subject testing.



Ojeda, M.; Lin J.; Ding, D.; (2012). Estimating Stroke Number and Cadence of Wheelchair Propulsion Using Portable Sensors. Proceedings of the 2012 RESNA Conference. Baltimore, MD.

Ding, D.; Hiremath, S.; Chung, Y.; Cooper, R. (2011). Detection of Wheelchair User Activities using Wearable Sensors. Lecture Notes in Computer Science, 6767, 145-152.


Return Home

This work is funded by the National Institute on Disability and Rehabilitation Research (NIDRR),
Rehabilitation Engineering Research Center (RERC) on Spinal Cord Injury, Grant #H133E070024
The ideas and opinions expressed herein are those of the authors and not necessarily reflective of the NIDRR.

Contact Webmaster | Accessibility Statement

Last Updated: 07.10.2012 | 16:55

Smart Code