Classifying physical activity levels using Mean Amplitude Deviation in adults using a chest worn accelerometer: validation of the Vivalink ECG Patch

Abstract

Abstract Background The development of readily available wearable accelerometers has enabled clinicians to objectively monitor physical activity (PA) remotely in the community, a superior alternative to patient self-reporting measures. Critical to the value of these monitors is the ability to reliably detect when patients are undergoing ambulatory activity. Previous studies have highlighted the strength of using mean amplitude deviation (MAD) as a universal measure for analysing raw accelerometery data and defining cut-points between sedentary and ambulatory activities. Currently however there is little evidence surrounding the use of chest-worn accelerometers which can provide simultaneous monitoring of other physiological parameters such as heart rate (HR), RR intervals, and Respiratory Rate alongside accelerometery data. We aimed to calibrate the accelerometery function within the VivaLink ECG patch to determine the cut-point MAD value for differentiating sedentary and ambulatory activities. Methods We recruited healthy volunteers to undergo a randomised series of 9 activities that simulate typical free-living behaviours, while wearing a VivaLink ECG Patch (Campbell, California). MAD values were applied to a Generalised Linear Mixed Model to determine cut-points between sedentary and ambulatory activities. We constructed a Receiver Operating Characteristic (ROC) curve to analyse the sensitivity and specificity of the cut-off MAD value. Results Eighteen healthy adults volunteered to the study and mean MAD values were collected for each activity. The optimal MAD cut-point between sedentary and ambulatory activities was 47.73mG. ROC curve analysis revealed an area under the curve of 0.99 (p < 0.001) for this value with a sensitivity and specificity of 98% and 100% respectively. Conclusion In conclusion, the MAD cut-point determined in our study is very effective at categorising sedentary and ambulatory activities among healthy adults and may be of use in monitoring PA in the community with minimal burden. It will also be useful for future studies aiming to simultaneously monitor PA with other physiological parameters via chest worn accelerometers.

Keywords

• Accelerometer
• Physical activity
• Mean amplitude deviation
• MAD
• Calibration
• Classification