The Merit of an Individual Calibration: Estimating Physical Activity Energy Expenditure in Wheelchair Users

INTRODUCTION: The use of multi-sensor devices, which integrate accelerometry and physiological signals to predict PAEE, is widespread in able-bodied humans. However, the validity of such devices remains to be established in wheelchair users. The aim of this study is to assess the error of a multi-sensor physical activity monitor to predict PAEE in wheelchair users and to assess the efficacy of individual heart rate calibration. METHODS: Seventeen manual wheelchair users (36±10 yrs, 72±11 kg) completed an activity of daily living and wheelchair propulsion at various velocities and gradients on a motorised wheelchair treadmill. Participants wore a combined accelerometer and heart rate monitor (ActiheartTM). Heart rate was also monitored during an incremental arm crank ergometer test to permit retrospective individual calibration of the ActiheartTM for the activity protocol. Linear regression analysis was conducted between criterion (indirect calorimetry) and estimated PAEE from the ActiheartTM using manufacturers proprietary algorithms or individual calibration. Subsequent error statistics were calculated. RESULTS: Predicted PAEE was significantly (p<0.01) associated with criterion PAEE (manufacturer algorithm; r=0.76 and individual calibration; r=0.95, Figure 1). Average PAEE estimation error for all activities was 14.6±63.2% and -10.1±20.7% using the manufacturer algorithm and individual calibration, respectively. Absolute PAEE estimation errors using the manufacturer algorithm varied from 42.4 to 66.3% and from 10.4 to 43.1% using individual calibration. DISCUSSION & CONCLUSION: The error associated with predicting PAEE in manual wheelchair users, using a multi-sensor device, is improved using an individual heart rate calibration. The inter-individual variance in cardiovascular response to exercise is high among this population. Therefore, we advocate the importance of using individual calibration of multi-sensor devices to predict PAEE in manual wheelchair users.