Field evaluation of a random forest activity classifier for wrist-worn accelerometer data

Pavey, Toby G., Gilson, Nicholas D., Gomersall, Sjaan R., Clark, Bronwyn and Trost, Stewart G. (2016) Field evaluation of a random forest activity classifier for wrist-worn accelerometer data. Journal of Science and Medicine in Sport, . doi:10.1016/j.jsams.2016.06.003

Author Pavey, Toby G.
Gilson, Nicholas D.
Gomersall, Sjaan R.
Clark, Bronwyn
Trost, Stewart G.
Title Field evaluation of a random forest activity classifier for wrist-worn accelerometer data
Journal name Journal of Science and Medicine in Sport   Check publisher's open access policy
ISSN 1878-1861
Publication date 2016
Year available 2016
Sub-type Article (original research)
DOI 10.1016/j.jsams.2016.06.003
Open Access Status Not Open Access
Total pages 6
Place of publication Chatswood, NSW Australia
Publisher Elsevier Australia
Collection year 2017
Language eng
Formatted abstract

Wrist-worn accelerometers are convenient to wear and associated with greater wear-time compliance. Previous work has generally relied on choreographed activity trials to train and test classification models. However, validity in free-living contexts is starting to emerge. Study aims were: (1) train and test a random forest activity classifier for wrist accelerometer data; and (2) determine if models trained on laboratory data perform well under free-living conditions.


Twenty-one participants (mean age = 27.6 ± 6.2) completed seven lab-based activity trials and a 24 h free-living trial (N = 16).


Participants wore a GENEActiv monitor on the non-dominant wrist. Classification models recognising four activity classes (sedentary, stationary+, walking, and running) were trained using time and frequency domain features extracted from 10-s non-overlapping windows. Model performance was evaluated using leave-one-out-cross-validation. Models were implemented using the randomForest package within R. Classifier accuracy during the 24 h free living trial was evaluated by calculating agreement with concurrently worn activPAL monitors.


Overall classification accuracy for the random forest algorithm was 92.7%. Recognition accuracy for sedentary, stationary+, walking, and running was 80.1%, 95.7%, 91.7%, and 93.7%, respectively for the laboratory protocol. Agreement with the activPAL data (stepping vs. non-stepping) during the 24 h free-living trial was excellent and, on average, exceeded 90%. The ICC for stepping time was 0.92 (95% CI = 0.75–0.97). However, sensitivity and positive predictive values were modest. Mean bias was 10.3 min/d (95% LOA = −46.0 to 25.4 min/d).


The random forest classifier for wrist accelerometer data yielded accurate group-level predictions under controlled conditions, but was less accurate at identifying stepping verse non-stepping behaviour in free living conditions Future studies should conduct more rigorous field-based evaluations using observation as a criterion measure.
Keyword Accelerometer
Random forest classifier
Physical activity
Q-Index Code C1
Q-Index Status Provisional Code
Institutional Status UQ

Document type: Journal Article
Sub-type: Article (original research)
Collections: HERDC Pre-Audit
School of Public Health Publications
School of Human Movement and Nutrition Sciences Publications
Version Filter Type
Citation counts: Scopus Citation Count Cited 0 times in Scopus Article
Google Scholar Search Google Scholar
Created: Tue, 02 Aug 2016, 15:56:39 EST by Sandrine Ducrot on behalf of School of Human Movement and Nutrition Sciences