Improving Hip-Worn Accelerometer Estimates of Sitting Using Machine Learning Methods

AbstractPurposeTo improve estimates of sitting time from hip worn accelerometers used in large cohort studies by employing machine learning methods developed on free living activPAL data.MethodsThirty breast cancer survivors concurrently wore a hip worn accelerometer and a thigh worn activPAL for 7 days. A random forest classifier, trained on the activPAL data, was employed to detect sitting, standing and sit-stand transitions in 5 second windows in the hip worn accelerometer. The classifier estimates were compared to the standard accelerometer cut point and significant differences across different bout lengths were investigated using mixed effect models.ResultsOverall, the algorithm predicted the postures with moderate accuracy (stepping 77%, standing 63%, sitting 67%, sit to stand 52% and stand to sit 51%). Daily level analyses indicated that errors in transition estimates were only occurring during sitting bouts of 2 minutes or less. The standard cut point was significantly different from the activPAL across all bout lengths, overestimating short bouts and underestimating long bouts.ConclusionsThis is among the first algorithms for sitting and standing for hip worn accelerometer data to be trained from entirely free living activPAL data. The new algorithm detected prolonged sitting which has been shown to be most detrimental to health. Further validation and training in larger cohorts is warranted.This is an open access article distributed under the Creative Commons Attribution License 4.0 (CCBY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Purpose To improve estimates of sitting time from hip worn accelerometers used in large cohort studies by employing machine learning methods developed on free living activPAL data. Methods Thirty breast cancer survivors concurrently wore a hip worn accelerometer and a thigh worn activPAL for 7 days. A random forest classifier, trained on the activPAL data, was employed to detect sitting, standing and sit-stand transitions in 5 second windows in the hip worn accelerometer. The classifier estimates were compared to the standard accelerometer cut point and significant differences across different bout lengths were investigated using mixed effect models. Results Overall, the algorithm predicted the postures with moderate accuracy (stepping 77%, standing 63%, sitting 67%, sit to stand 52% and stand to sit 51%). Daily level analyses indicated that errors in transition estimates were only occurring during sitting bouts of 2 minutes or less. The standard cut point was significantly different from the activPAL across all bout lengths, overestimating short bouts and underestimating long bouts. Conclusions This is among the first algorithms for sitting and standing for hip worn accelerometer data to be trained from entirely free living activPAL data. The new algorithm detected prolonged sitting which has been shown to be most detrimental to health. Further validation and training in larger cohorts is warranted. This is an open access article distributed under the Creative Commons Attribution License 4.0 (CCBY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Corresponding author: Jacqueline Kerr, UCSD, 9500 Gilman Dr, La Jolla, 92093-0810, Tel: 858 534 9305. Email: jkerr@ucsd.edu There were no conflicts of interest reported by authors of this study. This study was funded by pilot funding from the University of California, San Diego, Department of Family Medicine and Public Health. The University of California, San Diego, Department of Family Medicine and Public Health was not involved in the study design, data collection, analysis or submission for publication. The analyses and conclusions presented here are those of the authors and do not reflect those of the funders. In addition, the results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by American College of Sports Medicine. Accepted for Publication: 5 January 2018 © 2018 American College of Sports Medicine

from #ORL-AlexandrosSfakianakis via ola Kala on Inoreader http://ift.tt/2BXtnOR