Evaluating Simple Objective Metrics for the Remote Measurement of Physical Activity: Preliminary Results from the RADAR‐AD Study

Background Higher physical activity is associated with better global cognition and a lower risk of dementia (Rojer et al., 2021). Using questionnaires to assess physical activity is difficult in an Alzheimer’s disease (AD) population, since questionnaires rely on retrospective subjective recall. Wearables measure physical activity objectively and continuously. A variety exists, with some delivering raw high‐frequency research‐grade movement data, others providing summary variables only. The aim of this study is to compare simple activity metrics from two activity trackers monitoring physical activity in AD, and relate these to standard questionnaires. Method Participants in the RADAR‐AD study (Remote Assessment of Disease and Relapse – Alzheimer’s Disease) were asked to wear two activity trackers (dominant hand: Axivity AX3, non‐dominant hand: Fitbit Charge 3) for 8 weeks. Features calculated from the Axivity’s raw accelerometer data were: acceleration magnitude, time spent in sedentary, light, moderate‐to‐vigorous activity, and sleeping. Fitbit features included: step count and heart rate. Self‐reported activity levels (Godin Leisure Time Questionnaire), depression (Geriatric Depression Scale), social functioning (Social Functioning Scale), global cognition (Mini‐Mental State Examination), partner‐reported function (ADCS‐ADL) and neuropsychiatric symptoms (Neuropsychiatric Inventory) were assessed and related to activity levels using regression models. Result We included amyloid negative controls (n = 36) and amyloid positive preclinical (n = 13), prodromal (n = 15) and mild‐to‐moderate (n = 10) AD participants (Table 1). Expected associations were found between sedentary activity time and age, and between self‐reported activity and both moderate‐to‐vigorous activity time and step count (Table 2, Figure 1). Small group sizes currently make further comparisons of exploratory value only. Conclusion Wearable devices have the potential to accurately capture clinically relevant aspects of daily activity and function of interest in AD population. On‐going research in the RADAR‐AD study will focus on developing disease‐specific methods of feature extraction based on raw movement data, and combining features from multiple devices, to explore this further. This work has received support from the EU/EFPIA Innovative Medicines Initiative Joint Undertaking (grant No 806999). www.imi.europa.eu . This communication reflects the views of the RADAR-AD consortium and neither IMI nor the European Union and EFPIA are liable for any use that may be made of the information contained herein .