Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Using wearable devices to measure physical activity in manual wheelchair users with spinal cord injuries

Shwetar, Yousif (2020) Using wearable devices to measure physical activity in manual wheelchair users with spinal cord injuries. Undergraduate Thesis, University of Pittsburgh. (Unpublished)

Download (812kB) | Preview


Manual wheelchair users (MWUs) with spinal cord injury (SCI) generally exhibit low levels of physical activity (PA), placing them at a greater risk for many chronic diseases. Accurately measuring levels of PA in this population could potentially lead to better health management among these individuals. Recently, there has been a growth in the use of wearable devices to help individuals track free-living PA for self-management. This has been explored extensively in the ambulatory population, specifically with research grade activity monitors such as ActiGraph wearable devices. However, the literature lacks adequate investigation for energy expenditure (EE) assessment and PA estimation using wearable devices in the non-ambulatory population. The objective of this thesis is to assess the ability of wearable devices in estimating EE and PA in wheelchair users with SCI. In the first study, we conducted a literature search for existing EE predictive algorithms using an ActiGraph activity monitor for MWUs with SCI and evaluated their validity using an out-of-sample dataset collected from MWUs with chronic SCI. None of the five sets of predictive equations demonstrated equivalence within 20% of the criterion measure based on an equivalence test. The mean absolute error (MAE) for the five sets of predictive equations ranged from 0.87 – 6.41 kilocalories per minute (kcalmin-1) when compared with the criterion measure, and the intraclass correlation (ICC) estimates ranged from 0.06 – 0.59. Given the unsatisfactory performance of the existing EE predictive models, in the second study, we used machine learning techniques to develop a random forest model (RFM) for activity intensity estimation using data collected from MWUs with SCIs. Based on a 10-fold cross validation, the RFM had an average overall accuracy of 81.3% in distinguishing among sedentary, light-intensity PA, and MVPA with a precision of 0.82, 0.77, and 0.87, and a recall of 0.84, 0.79, and 0.82 for each intensity category, respectively. The results indicate that the RFM could classify sedentary and MVPA time reasonably well, but may lack the ability to classify light-intensity PA.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Shwetar, Yousifyjs4@pitt.eduyjs4
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairDing, DanDAD5@pitt.edudad5
Committee MemberCrytzer, Thereseatmc38@pitt.edutmc38
Committee MemberKoontz, Aliciaakoontz@pitt.eduakoontz
Committee MemberHiremath,
Date: 4 May 2020
Date Type: Publication
Defense Date: 16 April 2020
Approval Date: 4 May 2020
Submission Date: 16 April 2020
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 47
Institution: University of Pittsburgh
Schools and Programs: David C. Frederick Honors College
Swanson School of Engineering > Bioengineering
Degree: BSE - Bachelor of Science in Engineering
Thesis Type: Undergraduate Thesis
Refereed: Yes
Uncontrolled Keywords: Activity monitors, non-ambulatory, machine learning algorithms
Date Deposited: 04 May 2020 14:16
Last Modified: 04 May 2020 14:16


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item