Ann Geriatr Med Res.  2021 Jun;25(2):86-92. 10.4235/agmr.21.0034.

Comparison of Human Interpretation and a Rule-Based Algorithm for Instrumented Sit-to-Stand Test

Affiliations
  • 1Division of Geriatrics, Department of Internal Medicine, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
  • 2Dyphi Research Institute, Dyphi Inc., Daejeon, Korea

Abstract

Background
The five times sit-to-stand test (5STS) is one of the most commonly used tests to assess the physical performance of lower extremities. This study assessed the correlation between human interpretation (5STShuman) and a rule-based algorithm (5STSrule) using instrumented 5STS with two sensors.
Methods
We analyzed clinical records of 148 patients who visited the geriatric outpatient clinic of Asan Medical Center between December 2020 and March 2021 and underwent physical performance assessment using the electronic Short Physical Performance Battery (eSPPB) protocol. For STS, time-weight and time-distance curves were constructed using a loadcell and light detection and ranging (LiDAR). We manually assessed the grids of these curves to calculate 5STShuman, while 5STSrule used an empirical rule-based algorithm.
Results
In the study population, the mean 5STShuman and 5STSrule times, i.e., 12.2±0.4 and 11.4±0.4 seconds, respectively, did not differ significantly (p=0.232). Linear regression analysis showed that 5STShuman and 5STSrule were positively correlated (β=0.99, R2=0.99). The measures also did not differ (p=0.381) in classifying sarcopenia according to the Asian Working Group Society criteria, with C-indices of 0.826 for 5STShuman and 0.820 for 5STSrule.
Conclusion
An empirical rule-based algorithm correlated with human-interpreted 5STS and had comparable classification ability for sarcopenia.

Keyword

Sarcopenia; Physical performances; Aging; Biomarkers
Full Text Links
  • AGMR
Actions
Cited
CITED
export Copy
Close
Share
  • Twitter
  • Facebook
Similar articles
Copyright © 2022 by Korean Association of Medical Journal Editors. All rights reserved.     E-mail: koreamed@kamje.or.kr