Assessing Physical Activity in Manual Wheelchair Users

R1: Free-Living Physical Activity Assessment of Wheelchair Users Using Off-the-Shelf Activity Monitors




R1activity monitor

People who use wheelchairs for mobility tend to have lower physical activity levels than able-bodied populations. As a result of this physical inactivity, this population is associated with high incidences of chronic conditions and secondary complications including Type 2 diabetes, cardiovascular disease, fatigue, weight gain, pain, and depression. The availability of activity monitors can assist MWU with information to self-motivate and perform regular physical activity.

Energy expenditure (EE) monitoring devices, such as pedometers and accelerometers, are an effective way to provide people feedback regarding daily EE as an aid in attaining and maintaining healthy body weight. Unfortunately, there are no commercially available EE devices that are valid and reliable for assessing physical activity in manual wheelchair users (MWU). As a result, people who use wheelchairs as their primary mode of ambulation (e.g., spinal cord injury, cerebral palsy, spina bifida etc.) have no readily available technology to help calculate their daily EE in order to maintain an appropriate body weight or to lose weight.

Existing technology of wheel rotation data logger only measures physical activity in terms of velocity and distance traveled. New technology is needed in order to monitor upper body movements in manual wheelchair users and thereby enhance the accuracy of physical activity measurements. Current activity monitors are effective for capturing gross estimates of physical movements, but fall short when assessing physical activity in manual wheelchair users according to the type, intensity, or amount of physical activity in terms of energy expenditure across the day and are unable to measure physical activity performed outside of the wheelchair (hand cycling, wheelchair sports, etc.).

During the previous cycle of RecTech 4 activity monitors were evaluated, including senswear, RT3, actigraph, and wheel rotation data logger in MWC users with SCI. The study compared the output of these activity monitors to a portable metabolic cart, the gold standard. The study included 45 wheelchair users with spinal cord injury (SCI). Participants completed activites such as resting, propelling a wheelchair, performing arm ergometer exercises, and desk work. The results of the study showed large discrepancies between outputs of commercial devices and the gold standard metabolic cart. The estimation error ranged from 20-120%. The study showed that commercial devices are not suitable for MWU without further modifications.



Specific Aim 1: To develop prediction models of EE in MWUs during a variety of lifestyle and sporting activities for three popular off-the-shelf activity monitors.

Specific Aim 2: To validate the accuracy of the EE prediction models developed in Aim 1 in a separate sample of MWUs under semi-naturalistic and naturalistic conditions.

Specific Aim 3: To develop a web-based application for MWUs and open-source API for researchers to allow easy access to accurate PA and EE feedback from the three off-the-shelf activity monitors


Phase I entails the development of EE prediction models for MWUs based upon three popular off-the-shelf activity monitors. 63 subjects will visit site and be fitted with COSMED K4b2 portable telemetry system and all three activity monitors simultaneously prior to testing. Participants will be asked to perform 5 sets of activities including:

  1. Typical activities of daily living such as working on a computer, sitting and watching TV, reading, vacuuming, moving furniture/household items, doing laundry, and preparing meals
  2. Propelling their wheelchair at a self-selected comfortable pace, a fast pace, and up/down a slope
  3. Exercising with an arm ergometer at a self-selected comfortable pace and a fast pace
  4. Exercising with Thera-band and strength equipment
  5. Playing wheelchair basketball

Phase II entails the evaluation of the EE prediction models under semi-naturalistic and naturalistic conditions. Subjects will pay one visit to the study site (HERL or Lakeshore) and complete the same set of activities as described in Phase I. Subjects will also be tested at their homes where they will be asked to perform activities that are part of their daily routines for 2-3 hours.

Phase III entails the development of a web-based application and open-source API to allow MWUs and researchers to have access to accurate physical activity and EE information through the three activity monitors. The web-based interface will include the option for input of key demographic and anthropometric information required by EE prediction models, selection of the activity monitor they use, ability to upload data file collected from activity monitor. The application will also allow the user to view the activity summary including energy expenditure and the amount of time spent in sedentary, moderate, and vigorous physical activity.

Final Outcomes

R1 -
We expect the study to lead to tools that can facilitate better personal and clinical decisions about physical activity and energy balance and lead to healthier lifestyles for wheelchair users. This will be accomplished by packaging customized algorithms in web based applications and open sourced APIs that can be accessed by wheelchair users, clinicians, and researchers.

For more information on the EE_Estimator App, check out the products page.