Free-living dynamic skeletal loading estimation using tri-axial accelerometers

Hip bone mineral density (BMD) is an important biomarker of fracture risk, but as a standalone measure it lacks information on modifiable causal factors. The relationship between hip BMD and physical activity (PA)-based skeletal loading estimates could provide a biomarker for prescribing PA changes in high risk patients. Ten older women (age: 78.1 (8.1) yrs, BMI: 25.4 (3.8) kg/m-2) each performed 21 to 38 walking trials in the lab at slow, normal, and fast self-selected speeds (0.5-1.7 m/s) while wearing bilateral ankle activity monitors (AMs) (ActiGraph; 100 Hz). Ground reaction force (GRF) data were collected at 600 Hz for one to five steps per trial using five force plates and video data were collected at 60 Hz. Participants received DXA scans to measure total hip BMD and wore the AMs for 4 days in their free-living environments. Mean step count agreement was 94% with video and peak vertical GRF had a moderate significant correlation with acceleration (Fig 1). In this study, we observed a much smaller RMS error of 0.073 body weight (BW) for GRF estimations compared to 0.185 BW with the NASA GRF AM [1]. The peak vertical GRF of each step detected in the free-living environment was estimated from the peak impact vertical acceleration using a general GRF equation developed from the lab data (Fig 1). The mean (SD) daily step counts across participants were 10257 (3942). A bone density index (BDI) was calculated for each participant as the mean daily cumulative sum of each step’s estimated GRF. BMD had a high significant correlation with BDI. Data is currently being collected on an additional 70 participants for further algorithm development and validation. Future work will determine if this ankle acceleration based BDI tool can be used to predict BMD changes. [1] Bowley and Whalen, Ortho Res Soc, 2001.