Projects: Henrik Eckmann

Project Overview

My project aims to advance methods for estimating daily physical activity volume and intensity from wearable accelerometry, with a special focus on comparability of volume and intensity outcomes, and their associations with health, when derived from different common epoch summary measures.

First Study

The intensity gradient (IG) is a measure of the distribution of accelerometer-measured daily accelerations. Since its introduction in 2018, use of the IG in studies of cardiometabolic risk, waist-to-height-ratio and metabolic syndrome risk, incident cardiovascular disease, life expectancy, and all-cause mortality has improved our understanding of the role of the intensity distribution, independent of and in addition to the volume, of physical activity. It has been used in a variety of studies, usually with the Euclidean Norm Minus One (ENMO) as the acceleration epoch summary measure, in line with the original implementation, but sometimes applied to other epoch summary measures in use in accelerometer research, e.g. ActiGraph Counts. An epoch summary measure summarises raw accelerations, often tri-axial and sampled at up to 100 Hz, into a single acceleration value per a given length of time, called the epoch. Typical epoch lengths are five seconds to a minute. With increasing implementation of the IG in research, there is a clear need to assess the impacts of applying it to epoch summary measures other than ENMO, and how best to apply it, including potential necessary modifications to the algorithm. The aim of this study was to explore adaptations to the IG for use across other accelerometer epoch summary measures by 1) generating appropriate IG bins for each of MAD, MIMS, and Counts, and 2) assessing the agreement between the IGs generated using different epoch summary measures.

Second Study

Diverse studies implement diverse accelerometer measurement protocols. Accelerometers are commonly worn either on the wrist all 24 hours for the duration of the measurement, or they are worn at the waist during waking hours. How comparable measurements using either protocol are to measurements using the other is unclear, as is whether they may be more or less comparable depending on the epoch summary measure used. This study aims to compare PA volume and intensity measured 24-hours at either wrist and measured at the waist during waking hours for each of ENMO, MAD, MIMS, and Counts to assess both the overall comparability between protocols, and to discern if any epoch summary measure may be more or less sensitive to such protocol differences.

Third Study

Diverse studies implement diverse accelerometer measurement protocols. Accelerometers are commonly worn either on the wrist all 24 hours for the duration of the measurement, or they are worn at the waist during waking hours. How comparable measurements using either protocol are to measurements using the other is unclear, as is whether they may be more or less comparable depending on the epoch summary measure used. This study aims to compare PA volume and intensity measured 24-hours at either wrist and measured at the waist during waking hours for each of ENMO, MAD, MIMS, and Counts to assess both the overall comparability between protocols, and to discern if any epoch summary measure may be more or less sensitive to such protocol differences.

Fourth Study

Use of device-based measurement of physical activity (PA) using wearable accelerometers has increased rapidly in PA and health research over the last couple of decades. This includes clinical studies, where device-based measures have potential to improve patient PA assessment, while keeping patient and clinician burden low. The SafeHeart study aims to develop an algorithm, integrated into a web-based clinician’s dashboard, that uses high-resolution accelerometer data to predict implanted cardioverter-defibrillator (ICD) therapy events. To maximise the clinical utility of device-based PA measurement for prediction of ICD therapy events, it is important to understand whether and how PA in patients with an ICD is associated with ICD therapy events. To accurately assess associations between daily PA and ICD therapy events, other sources impacting daily PA must also be understood. This includes the lived environment. For example, PA has been shown to vary with season, weather, weekends and holidays in general populations. Similarly, PA was shown to vary with season in one study of ICD patients. Assessment of how PA volume and intensity vary across season, weather, weekends/holidays, and whether they are associated with ICD events, will enhance our understanding of the PA of ICD patients and provide insight into the utility of PA volume and intensity, and specifically AvAcc and IG, for prediction of ICD therapy events. Additionally, comparing results when based on different epoch summary measures (ENMO, MAD, MIMS, Counts) will let us assess the comparability of analyses across epoch summary measures from yet another perspective. Therefore, this study aimed to assess 1) associations between the lived environment (weather, season, weekday/weekend/holiday) and wearable accelerometer-derived daily PA volume and PA intensity, 2) associations between proximity to ICD event and wearable accelerometer-derived daily PA volume and intensity, and 3) comparability of results between common epoch summary measures.

Fifth Study

The Norwegian National Physical Activity Survey (NNPAS) periodically measures the PA of a nationally representative sample of Norwegian adults using wearable accelerometry. Nationally representative age- and sex-referenced centiles of PA volume and intensity could serve as a useful reference point for future studies or clinicians assessing PA volume and intensity, and feed into national PA policy. Creating reference centiles based on all of ENMO, MAD, MIMS, and Counts will allow a wide audience to compare they’re results and will provide the basis for further comparison between epoch summary measures. This study aims to 1) generate age- and sex-referenced centiles of PA volume (AvAcc) and intensity (IG) from ENMO, MAD, MIMS, and Counts, and 2) compare centile ranks between epoch summary measures.