Automatic scoring of chair sit-to-stand test using a smartphone

Sher, A ORCID logoORCID: https://orcid.org/0000-0002-0650-0335, Langford, D, Villagra, F and Akanyeti, O, 2024. Automatic scoring of chair sit-to-stand test using a smartphone. In: Panoutsos, G, Mahfouf, M and Mihaylova, LS, eds., Advances in computational intelligence systems: contributions presented at the 21st UK Workshop on Computational Intelligence, September 7–9, 2022, Sheffield, UK. Advances in Intelligent Systems and Computing (1454). Cham: Springer, pp. 170-180. ISBN 9783031555671

[thumbnail of 2231187_Sher.pdf] Text
2231187_Sher.pdf - Post-print
Full-text access embargoed until 18 May 2025.

Download (1MB)

Abstract

Chair sit to stand tests (CST) are widely used in clinical settings to measure endurance, balance and lower extremity muscle strength. It also allows clinicians to predict falls and cognitive decline in older adults. The current CST measurements are done manually using a timer. The manual CST measurements can be imprecise (often leading to high inter-rater variability), and they ignore what kinematics strategies participants use to stand up and sit back on the chair. In this study, we present a smartphone based automatic CST analysis system. The system has the ability of generating a CST score, and perform cycle by cycle motion analysis. To achieve this, it employs two XGBoost classifiers one for recognising who is taking the test and which chair rising strategy they use. This information is then used to adapt its algorithms for more accurate CST score predictions. The performance of the system was tested on 30 participants including three demographics group (healthy young, healthy adult and Parkinson’s) who were using two different chair rising strategies (flexion and momentum transfer). Overall, the system had above 95% classification accuracy, and the mean absolute difference between predicted and actual CST cycle completion time was less than 60 ms (<10% considering that the average cycle completion time was above 1 s). These results are highly encouraging towards developing a new smartphone-based gait and balance assessment tool that can be used in outdoor settings.

Item Type: Chapter in book
Description: Paper presented at the 21st UK Workshop on Computational Intelligence, Sheffield, 7-9 September 2022.
Creators: Sher, A., Langford, D., Villagra, F. and Akanyeti, O.
Publisher: Springer
Place of Publication: Cham
Date: 2024
Number: 1454
ISBN: 9783031555671
Identifiers:
Number
Type
10.1007/978-3-031-55568-8_14
DOI
2231187
Other
Divisions: Schools > School of Science and Technology
Record created by: Jonathan Gallacher
Date Added: 30 Sep 2024 14:51
Last Modified: 30 Sep 2024 14:51
URI: https://irep.ntu.ac.uk/id/eprint/52324

Actions (login required)

Edit View Edit View

Statistics

Views

Views per month over past year

Downloads

Downloads per month over past year