Validation of seat-off and seat-on in repeated sit-to-stand movements using a single-body-fixed sensor

• Published in: Physiological measurement Vol. 33; no. 11; p. 1855
• Main Authors: van Lummel, R C, Ainsworth, E, Hausdorff, J M, Lindemann, U, Beek, P J, van Dieën, J H
• Format: Journal Article
• Language: English
• Published: England 01-11-2012
• Subjects: Accelerometry - instrumentation; Aged, 80 and over; Humans; Movement - physiology; Posture; Reproducibility of Results; Young Adult
• ISSN: 1361-6579
• DOI: 10.1088/0967-3334/33/11/1855

The identification of chair rise phases is a prerequisite for quantifying sit-to-stand movements. The aim of this study is to validate seat-off and seat-on detection using a single-body-fixed sensor against detection based on chair switches. A single sensor system with three accelerometers and three gyroscopes was fixed around the waist. Synchronized on-off switches were placed under the chair. Thirteen older adults were recruited from a residential care home and fifteen young adults were recruited among college students. Subjects were asked to complete two sets of five trials each. Six features of the trunk movement during seat-off and seat-on were calculated automatically, and a model was developed to predict the moment of seat-off and seat-on transitions. The predictions were validated with leave-one-out cross-validation. Feature extraction failed in two trials (0.7%). For the optimal combination of seat-off predictors, cross-validation yielded a mean error of 0 ms and a mean absolute error of 51 ms. For the best seat-on predictor, cross-validation yielded a mean error of -3 ms and a mean absolute error of 127 ms. The results of this study demonstrate that seat-off and seat-on in repeated sit-to-stand movements can be detected semi-automatically in young and older adults using a one-body-fixed sensor system with an accuracy of 51 and 127 ms, respectively. The use of the ambulatory instrumentation is feasible for non-technically trained personnel. This is an important step in the development of an automated method for the quantification of sit-to-stand movements in clinical practice.