Identification of meaningful and sensitive metrics to capture the different dimensions of physical activity using 24h time-series accelerometry data

Aim

Our team aims at identifying metrics that have been developed so far to define physical activity patterns using raw accelerometry data. There is a strong need for a scoping review on the available approaches for the identification of physical activity patterns.

These methods may be used in future studies on the association between physical activity and health conditions (e.g. cardiometabolic health, Diabetes, frailty), as well as in personalised interventions in public health. This work will also provide consumer wearable device companies with decisive information on future developments in the data processing, as well as on relevant feedback to the end-user.

Background

Current knowledge on how health behaviours such as physical activity, sitting and sleep affect our health is based on self-report by questionnaires, which have limited validity, are prone to bias and enquire about selective aspects of these behaviours.

Progresses in technology (e.g. accelerometry) and methods (e.g. compositional analysis, acceleration distribution) have changed the landscape. One of the most exciting aspects of accelerometers is that they theoretically allow for capturing nearly complete accounts of movement behaviour, including posture and activity type detection. Interest in 24-hour accelerometry to assess all movement behaviours is increasing but analyses are complicated. Actually, only few studies have used the devices collecting data in raw mode 24/7 to cover activity pattern of the whole day.

Most accelerometer-based constructs of physical activity are based on total daily time spent in different intensity bands. However, ongoing advancements in accelerometers and data science have opened new avenues for incorporating a variety of other equally important characteristics of physical activity, like posture, activity types and sleep.

 

 

These new approaches will allow to:

1) define multidimensional profiles constructed across the key physical activity dimensions,

2) investigate the association between profiles and health outcomes,

3) personalise future interventions in public health.