Device-assessed physical activity and sedentary behavior in a community-based cohort of older adults.

BACKGROUND: Few studies characterize older adult physical activity and sitting patterns using accurate accelerometer and concurrent posture measures. In this descriptive paper, we report accelerometer data collection protocols, consent rates, and physical behavior measures from a population-based cohort study (Adult Changes in Thought, ACT). METHODS: The ACT study holds enrollment steady at approximately 2000 members of Kaiser Permanente Washington aged 65+ without dementia undergoing detailed biennial assessments. In 2016 the ACT-Activity Monitor (ACT-AM) sub-study was initiated to obtain data from wearing activPAL and ActiGraph devices for 7 days following regular biennial visits. We describe the methods protocol of ACT-AM and present characteristics of people who did and did not consent to wear devices. We compute inverse probability of response weights and incorporate these weights in linear regression models to estimate means and 95% confidence intervals (CI) of device-based pattern metrics, adjusted for wear time and demographic factors, and weighted to account for potential selection bias due to device-wear consent. RESULTS: Among 1885 eligible ACT participants, 56% agreed to wear both devices (mean age 77 years, 56% female, 89% non-Hispanic white, 91% with post-secondary education). On average, those who agreed to wear devices were younger and healthier. Estimated mean (95% CI) activPAL-derived sitting, standing, and stepping times were 10.2 h/day (603-618 min/day), 3.9 h/day (226-239 min/day), and 1.4 h/day (79-84 min/day), respectively. Estimated mean ActiGraph derived sedentary (Vector Magnitude [VM] < =18 counts/15 s), light intensity (VM 19-518 counts/15 s), and moderate-to-vigorous intensity (VM > 518 counts/15 s) physical activity durations were 9.5 h/day (565-577 min/day), 4.5 h/day (267-276 min/day), and 1.0 h/day (59-64 min/day). Participants who were older, had chronic conditions, and were unable to walk a half-mile had higher sedentary time and less physical activity. CONCLUSIONS: Our recruitment rate demonstrates the feasibility of cohort participants to wear two devices that measure sedentary time and physical activity. Data indicate high levels of sitting time in older adults but also high levels of physical activity using cut-points developed for older adults. These data will help researchers test hypotheses related to physical behavior and health in older adults in the future.

A Kernel-based Test of Independence for Cluster-correlated Data.

The Hilbert-Schmidt Independence Criterion (HSIC) is a powerful kernel-based statistic for assessing the generalized dependence between two multivariate variables. However, independence testing based on the HSIC is not directly possible for cluster-correlated data. Such a correlation pattern among the observations arises in many practical situations, e.g., family-based and longitudinal data, and requires proper accommodation. Therefore, we propose a novel HSIC-based independence test to evaluate the dependence between two multivariate variables based on cluster-correlated data. Using the previously proposed empirical HSIC as our test statistic, we derive its asymptotic distribution under the null hypothesis of independence between the two variables but in the presence of sample correlation. Based on both simulation studies and real data analysis, we show that, with clustered data, our approach effectively controls type I error and has a higher statistical power than competing methods.

A Flexible Framework for Nonparametric Graphical Modeling that Accommodates Machine Learning.

Graphical modeling has been broadly useful for exploring the dependence structure among features in a dataset. However, the strength of graphical modeling hinges on our ability to encode and estimate conditional dependencies. In particular, commonly used measures such as partial correlation are only meaningful under strongly parametric (in this case, multivariate Gaussian) assumptions. These assumptions are unverifiable, and there is often little reason to believe they hold in practice. In this paper, we instead consider 3 nonparametric measures of conditional dependence. These measures are meaningful without structural assumptions on the multivariate distribution of the data. In addition, we show that for 2 of these measures there are simple, strong plug-in estimators that require only the estimation of a conditional mean. These plug-in estimators (1) are asymptotically linear and non-parametrically efficient, (2) allow incorporation of flexible machine learning techniques for conditional mean estimation, and (3) enable the construction of valid Wald-type confidence intervals. In addition, by leveraging the influence function of these estimators, one can obtain intervals with simultaneous coverage guarantees for all pairs of features.