Reallocation of time between device-measured movement behaviours and risk of incident cardiovascular disease

Abstract

Objective: To improve classification of movement behaviours in free-living accelerometer data using machine-learning methods, and to investigate the association between machine-learned movement behaviours and risk of incident cardiovascular disease (CVD) in adults.

Methods: Using free-living data from 152 participants, we developed a machine-learning model to classify movement behaviours (moderate-to-vigorous physical activity behaviours (MVPA), light physical activity behaviours, sedentary behaviour, sleep) in wrist-worn accelerometer data. Participants in UK Biobank, a prospective cohort, were asked to wear an accelerometer for 7 days, and we applied our machine-learning model to classify their movement behaviours. Using compositional data analysis Cox regression, we investigated how reallocating time between movement behaviours was associated with CVD incidence.

Results: In leave-one-participant-out analysis, our machine-learning method classified free-living movement behaviours with mean accuracy 88% (95% CI 87% to 89%) and Cohen's kappa 0.80 (95% CI 0.79 to 0.82). Among 87 498 UK Biobank participants, there were 4105 incident CVD events. Reallocating time from any behaviour to MVPA, or reallocating time from sedentary behaviour to any behaviour, was associated with lower CVD risk. For an average individual, reallocating 20 min/day to MVPA from all other behaviours proportionally was associated with 9% (95% CI 7% to 10%) lower risk, while reallocating 1 hour/day to sedentary behaviour from all other behaviours proportionally was associated with 5% (95% CI 3% to 7%) higher risk.

Conclusion: Machine-learning methods classified movement behaviours accurately in free-living accelerometer data. Reallocating time from other behaviours to MVPA, and from sedentary behaviour to other behaviours, was associated with lower risk of incident CVD, and should be promoted by interventions and guidelines.

Keywords: cardiovascular diseases; methods; physical activity; sedentary behavior; sleep.

Figure 1

Participant flow diagram for the analysis of movement behaviours and incident cardiovascular disease in UK Biobank participants.BMI, Body Mass Index; TDI, Townsend Deprivation Index.

Figure 2

Distribution of movement behaviours in 87 498 UK Biobank participants. (A) Mean movement behaviour composition among UK Biobank participants. (B) Movement behaviours of UK Biobank participants on a ternary plot, showing sleep, sedentary behaviour (SB) and physical activity behaviours (PA; combines light and moderate-to-vigorous physical activity behaviours). The crosshair marks the compositional mean. Concentric rings represent the 25, 50% and 75% prediction regions for the data. The behaviour composition at a point can be found by tracing out (parallel to the white lines and crosshair) from the point to the axes. (C) Ternary plot showing the behaviour distribution of the 5% most active (blue) and 5% least active (red) UK Biobank participants by average acceleration. Concentric rings represent the 25, 50% and 75% prediction regions for each group. LIPA, light physical activity behaviours; MVPA, moderate-to-vigorous physical activity behaviours.

Figure 3

HRs for incident cardiovascular disease associated with balance between movement behaviours in 87 498 UK Biobank participants.Model based on 4105 events in 87 498 participants. All relative to the mean behaviour composition (8.8 hours/day sleep, 9.3 hours/day sedentary behaviour (SB), 5.6 hours/day light physical activity behaviours (LIPA), 0.35 hours/day (21 min/day) moderate-to-vigorous physical activity behaviours (MVPA)). Model used age as the timescale, was stratified by sex and was additionally adjusted for ethnicity, smoking status, alcohol consumption, fresh fruit and vegetable consumption, red and processed meat consumption, oily fish consumption, deprivation and education. 95% CIs shown.

Figure 4

HRs for incident cardiovascular disease associated with reallocating time to named behaviour, from all other behaviours proportionally, in 87 498 UK Biobank participants.Model based on 4105 events in 87 498 participants. All relative to the mean behaviour composition (8.8 hours/day sleep, 9.3 hours/day sedentary behaviour (SB), 5.6 hours/day light physical activity behaviours (LIPA), 0.35 hours/day (21 min/day) moderate-to-vigorous physical activity behaviours (MVPA)) and more time in named behaviour reallocated from all other behaviours proportionally. Model used age as the timescale, was stratified by sex and was additionally adjusted for ethnicity, smoking status, alcohol consumption, fresh fruit and vegetable consumption, red and processed meat consumption, oily fish consumption, deprivation and education. 95% CIs shown.