Fixed-effects inference and tests of correlation for longitudinal functional data.

We propose an inferential framework for fixed effects in longitudinal functional models and introduce tests for the correlation structures induced by the longitudinal sampling procedure. The framework provides a natural extension of standard longitudinal correlation models for scalar observations to functional observations. Using simulation studies, we compare fixed effects estimation under correctly and incorrectly specified correlation structures and also test the longitudinal correlation structure. Finally, we apply the proposed methods to a longitudinal functional dataset on physical activity. The computer code for the proposed method is available at https://github.com/rli20ST758/FILF.

Classification of human physical activity based on raw accelerometry data via spherical coordinate transformation.

Human health is strongly associated with person's lifestyle and levels of physical activity. Therefore, characterization of daily human activity is an important task. Accelerometers have been used to obtain precise measurements of body acceleration. Wearable accelerometers collect data as a three-dimensional time series with frequencies up to 100 Hz. Using such accelerometry signal, we are able to classify different types of physical activity. In our work, we present a novel procedure for physical activity classification based on the raw accelerometry signal. Our proposal is based on the spherical representation of the data. We classify four activity types: resting, upper body activities (sitting), upper body activities (standing), and lower body activities. The classifier is constructed using decision trees with extracted features consisting of spherical coordinates summary statistics, moving averages of the radius and the angles, radius variance, and spherical variance. The classification accuracy of our method has been tested on data collected on a sample of 47 elderly individuals who performed a series of activities in laboratory settings. The achieved classification accuracy is over 90% when the subject-specific data are used and 84% when the group data are used. Main contributor to the classification accuracy is the angular part of the collected signal, especially spherical variance. To the best of our knowledge, spherical variance has never been previously used in the analysis of the raw accelerometry data. Its major advantage over other angular measures is its invariance to the accelerometer location shifts.

Deep Learning for Activity Recognition in Older People Using a Pocket-Worn Smartphone.

Activity recognition can provide useful information about an older individual's activity level and encourage older people to become more active to live longer in good health. This study aimed to develop an activity recognition algorithm for smartphone accelerometry data of older people. Deep learning algorithms, including convolutional neural network (CNN) and long short-term memory (LSTM), were evaluated in this study. Smartphone accelerometry data of free-living activities, performed by 53 older people (83.8 ± 3.8 years; 38 male) under standardized circumstances, were classified into lying, sitting, standing, transition, walking, walking upstairs, and walking downstairs. A 1D CNN, a multichannel CNN, a CNN-LSTM, and a multichannel CNN-LSTM model were tested. The models were compared on accuracy and computational efficiency. Results show that the multichannel CNN-LSTM model achieved the best classification results, with an 81.1% accuracy and an acceptable model and time complexity. Specifically, the accuracy was 67.0% for lying, 70.7% for sitting, 88.4% for standing, 78.2% for transitions, 88.7% for walking, 65.7% for walking downstairs, and 68.7% for walking upstairs. The findings indicated that the multichannel CNN-LSTM model was feasible for smartphone-based activity recognition in older people.

Low-cost wearable data acquisition for stroke rehabilitation: a proof-of-concept study on accelerometry for functional task assessment.

BACKGROUND: An increasingly aging society and consequently rising number of patients with poststroke-related neurological dysfunctions are forcing the rehabilitation field to adapt to ever-growing demands. Although clinical reasoning within rehabilitation is dependent on patient movement performance analysis, current strategies for monitoring rehabilitation progress are based on subjective time-consuming assessment scales, not often applied. Therefore, a need exists for efficient nonsubjective monitoring methods. Wearable monitoring devices are rapidly becoming a recognized option in rehabilitation for quantitative measures. Developments in sensors, embedded technology, and smart textile are driving rehabilitation to adopt an objective, seamless, efficient, and cost-effective delivery system. This study aims to assist physiotherapists' clinical reasoning process through the incorporation of accelerometers as part of an electronic data acquisition system. METHODS: A simple, low-cost, wearable device for poststroke rehabilitation progress monitoring was developed based on commercially available inertial sensors. Accelerometry data acquisition was performed for 4 first-time poststroke patients during a reach-press-return task. RESULTS: Preliminary studies revealed acceleration profiles of stroke patients through which it is possible to quantitatively assess the functional movement, identify compensatory strategies, and help define proper movement. CONCLUSION: An inertial data acquisition system was designed and developed as a low-cost option for monitoring rehabilitation. The device seeks to ease the data-gathering process by physiotherapists to complement current practices with accelerometry profiles and aid the development of quantifiable methodologies and protocols.

Extracting actigraphy-based walking features with structured functional principal components.

Objective.We extract walking features from raw accelerometry data while accounting for varying cadence and commonality of features among subjects. Walking is the most performed type of physical activity. Thus, we explore if an individual's physical health is related to these walking features.Approach.We use data collected using ActiGraph GT3X+ devices (sampling rate = 80 Hz) as part of the developmental epidemiologic cohort study,I= 48, age =78.7±5.7years, 45.8% women. We apply structured functional principal component analysis (SFPCA) to extract features from walking signals on both, the subject-specific and the subject-spectrum-specific level of a fast-paced 400 m walk, an indicator of aerobic fitness in older adults. We also use the subject-specific level feature scores to study their associations with age and physical performance measures. Specifically, we transform the raw data into the frequency domain by applying local Fast Fourier Transform to obtain the walking spectra. SFPCA decomposes these spectra into easily interpretable walking features expressed as cadence and acceleration, which can be related to physical performance.Main results.We found that five subject-specific and 19 subject-spectrum-specific level features explained more than 85% of their respective level variation, thus significantly reducing the complexity of the data. Our results show that 54% of the total data variation arises at the subject-specific and 46% at the subject-spectrum-specific level. Moreover, we found that higher acceleration magnitude at the cadence was associated with younger age, lower BMI, faster average cadence and higher short physical performance battery scores. Lower acceleration magnitude at the cadence and higher acceleration magnitude at cadence multiples 2.5 and 3.5 are related to older age and higher blood pressure.Significance.SFPCA extracted subject-specific level empirical walking features which were meaningfully associated with several health indicators and younger age. Thus, an individual's walking pattern could shed light on subclinical stages of somatic diseases.

Gender difference in "second-shift" physical activity: New insights from analyzing accelerometry data in a nationally representative sample.

The gendered organization of daily activities results in differential contexts of physical activity (PA) for the working population, especially during the "second shift" - a time window dominated by household-based activities. Existing research predominantly relies on self-reported leisure-time activities, yielding a partial understanding of gender difference in the source, timing, and accumulation pattern of PA. To address these limitations, this study draws on the interplay between work and family to understand how they shape gender difference in household-based PA across occupational groups. It combines work schedule and accelerometry PA data from the 2005-2006 National Health and Nutrition Examination Survey (NHANES), which permits our study of second-shift PA on workdays among full-time workers, aged 20 to 49, with a regular daytime schedule. To capture different aspects of second-shift PA, the PA outcomes are measured as both volume and accumulation patterns during time windows following (i.e., 6pm-9pm) and prior to typical working hours (7:30am-8:30am). Using generalized estimating equations, we estimate gender differences in the volume and fragmentation of second-shift PA. Overall, women with a full-time job exhibit both higher volume and higher fragmentation of second-shift PA than their male counterparts. The occupational group moderates such gender difference in PA. The gender gaps in PA volume and fragmentation are only evident for professional workers, whereas the second shift represents a gender-neutral context for PA accumulation for non-professional groups. These findings are supported by a secondary analysis when analyzing the whole-day PA data using functional data analysis. Such social patterning of second-shift PA calls for further research on gendered PA under the interplay of work and family beyond the usual focus on leisure activities.

Reference Values for Wristband Accelerometry Data in Children Aged 6-11 Years of Age.

Objectives: Wristband activity trackers (accelerometers) could serve as a convenient monitoring tool to continuously quantify physical activity throughout the day. We aim to provide reference values for the use of these devices in healthy children. Methods: Children were recruited at a local school and provided with activity trackers (Fitbit Charge 2). Pupils were instructed to wear devices during all normal daytime activities over a period of 11-15 days. Demographic data, total number of daily steps and heart rate were recorded. In addition, all children/parents were asked to complete a questionnaire providing information about daily physical routine (mode of transport to school, sporting activities as well as sport club memberships). Results: Three hundred two children (54.6% boys; median age 8.7 years) participated in this prospective study. Median wearing time of the device was 12.1 h/day. Overall, the median daily total step count was 12,095. Median step counts/day were significantly higher in boys compared to girls (13,015 vs. 11,305 steps/day; p < 0.0001). In addition, step counts were significantly higher during the week, compared to weekend days. The effect of age on daily step count was found to be non-linear: the total daily step count increased from 6 to 8.5 years of age, while older children (aged >8.5 years) had lower step counts compared to the younger children. Significant predictors of the daily step count were male gender (+1,324.9 steps, p = 0.0008), mode of transportation to school (walking, bicycle, scooter: +865.5 steps p = 0.049), active membership in a sports club (+1,324.9 steps/day, p = 0.0008), and number of structured units of physical exercise performed (+336.5/per 45 min, p < 0.0001). Severe obesity was associated with a significant reduction in total daily step count (-3037.7 steps/day, p = 0.015). Conclusion: Our prospective cohort study of healthy school children provides reference values for wristband accelerometers in normal individuals. In addition, it clarifies the effect of age, body weight and lifestyle on normal daily step counts in school children. This data should be helpful to judge the degree of physical limitation of patients compared to healthy peers.

Dataset of acceleration signals recorded while performing activities of daily living.

Several research studies have investigated the human activity recognition (HAR) domain to detect and recognise patterns of daily human activities. However, the accurate and automatic assessment of activities of daily living (ADLs) through machine learning algorithms is still a challenge, especially due to limited availability of realistic datasets to train and test such algorithms. The dataset contains data from 52 participants in total (26 women, and 26 men). The data for these participants was collected in two phases: 33 participants initially, and 19 further participants later on. Participants performed up to 5 repetitions of 24 different ADLs. Firstly, we provide an annotated description of the dataset collected by wearing a wrist-worn measurement device, Empatica E4. Secondly, we describe the methodology of the data collection and the real context in which participants performed the selected activities. Finally, we present some examples of recent and relevant target applications where our dataset can be used, namely lifelogging, behavioural analysis and measurement device evaluation. The authors consider the dissemination of this dataset can highly benefit the research community, and specially those involved in the recognition of ADLs, and/or in the removal of cues that reveal identity.