BACKGROUND:Understanding the association between movement behaviors and psychosocial well-being at a young age seems essential for effective interventions and moving friendly environments, particularly in the context of childhood obesity. OBJECTIVES:to examine the association between adherence to movement behavior guidelines and psychosocial well-being, and whether obesity indices moderate this association. METHODS:The IDEFICS/IFamily cohort followed European children aged 2-16 years over 6 years, including 7359 repeated observations across three waves. Longitudinal associations between adherence to movement behavior recommendations and psychological well-being were assessed using generalized linear models. Obesity indicators, z-score body mass index (z-BMI) and z-score waist circumference (z-BMI) were used to test their moderating role. RESULTS:Lower adherence to movement behaviors was negatively associated with psychosocial well-being (β = -0.39, 95%CI: -0.77, 0.00), with stronger effects in males (β = -0.70, 95%CI: -1.20, -0.20). Moreover, effects were larger in participants with overweight/obesity (β = -1.29, 95%CI: -2.21, -0.37). Obesity indices moderated the association between movement behaviors and psychosocial well-being (β = -0.40, 95%CI: -0.80, -0.06 for z-BMI; β = -0.36, 95%CI: -0.82, -0.03 for z-WC). CONCLUSIONS:Adherence to movement behavior guidelines was associated with psychosocial well-being, especially in males and individuals with overweight or obesity. These findings highlight the potential relevance of movement behavior patterns for psychosocial well-being and obesity prevention, while acknowledging the observational nature of the data.

Background:Accurate assessment of physical behaviors (PBs) and activity intensity is essential for public health research and digital health monitoring. Wearable accelerometers combined with machine learning (ML) or deep learning (DL) enable objective behavior assessment, but most existing models are trained on laboratory data, limiting generalizability to free-living conditions. Objective:This study aimed to develop and evaluate multitask ML and DL models for PB classification across 7 categories (sitting, standing, walking, running, sports, cycling, and lying) and activity intensity categories (AIC) across 3 levels (sedentary, light, and moderate-to-vigorous physical activity) using thigh-worn (activPAL) and waist-worn (ActiGraph) wearable accelerometers. A second objective was to compare model-derived estimates of daily time spent in PB and AIC across single- and dual-sensor (activPAL + ActiGraph) configurations, and to evaluate agreement between the best-performing model and corresponding estimates obtained from the proprietary activPAL classification of real-world everyday activities (CREA) algorithm using free-living data collected over a 9-day monitoring period. Methods:Data were obtained from 590 adults in the multicenter WEALTH study (627 recruited) and included up to 9 days of concurrent activPAL and ActiGraph free-living recordings. Sparse accelerometer-labeled data were obtained using ecological momentary assessment and refined by retaining instances with ≥75% agreement with the CREA algorithm. Resulting labeled data of 583 participants were used to develop ML models for single-sensor (activPAL or ActiGraph) and combined (dual-sensor) configurations. A random forest (RF) model using engineered features and a multihead convolutional neural network (MH-CNN) were trained within a multitask learning framework to jointly predict PB (task 1) and AIC (task 2) using a subject-independent hold-out split. The test subset (n=87) was used to estimate daily time spent in PB and AIC over 9 days, which were compared with CREA-derived estimates using Pearson coefficients and intraclass correlation coefficients (ICCs). Results:The dual-sensor configuration consistently outperformed single-sensor models. For PB classification, the MH-CNN achieved the highest performance (F1-score=0.750). For AIC, the RF model performed best (F1-score=0.741). Dual-sensor free-living estimates showed epidemiologically plausible distributions across the 24-hour period, including sitting 37% (538/1440 min), lying 34% (496/1440 min), walking 9% (131/1440 min), and moderate-to-vigorous physical activity (MVPA) 2% (31/1440 min). Agreement with CREA was strongest for standing, walking, and cycling (r≥0.86; ICC ≥0.72), while lying showed modest reliability (ICC=0.48). For AIC, agreement was highest for light physical activity (LPA) and MVPA (ICC 0.72-0.75). Conclusions:Multitask models combining thigh- and waist-worn accelerometers provide consistent estimates of PB and AIC under free-living conditions. The dual-sensor approach yielded more stable and epidemiologically coherent estimates than single-sensor methods, supporting its potential for large-scale population monitoring and mobile health apps.

Accurate assessment of physical behaviours (PB) and energy expenditure (EE) is essential for public health research and digital health monitoring. Wearable accelerometers combined with machine learning (ML) or deep learning (DL) enable objective behaviour assessment, but most existing models are trained on laboratory data, limiting generalisability to free-living conditions. This study aimed to develop and validate multitask ML and DL models for PB classification across seven categories (sitting, standing, walking, running, sports, cycling, and lying) and EE across three intensity levels (sedentary, light, and moderate-to-vigorous physical activity) using thigh-worn (activPAL) and waist-worn (ActiGraph) wearable accelerometers. A second objective was to compare model-derived estimates of daily time spent in PB and EE categories across single- and dual-sensor (activPAL+ActiGraph) configurations, and to evaluate agreement between the best-performing model and corresponding estimates obtained from the proprietary activPAL CREA algorithm using free-living data collected over a 9-day monitoring period. Data were obtained from 590 adults in the multicentre WEALTH study (627 recruited) and included up to 9 days of concurrent activPAL and ActiGraph free-living recordings. Sparse accelerometer-labelled data were obtained using ecological momentary assessment and refined by retaining instances with ≥75% agreement with the CREA algorithm. Resulting labelled data of 583 participants were used to develop ML models for single-sensor (activPAL or ActiGraph) and combined (dual-sensor) configurations. A random forest (RF) model using engineered features and a multi-head convolutional neural network (MH-CNN) were trained within a multitask learning framework to jointly predict PB (task 1) and EE (task 2) using a subject-independent hold-out split. The test subset (n=87) was used to estimate daily time spent in PB and EE categories over 9 days, which were compared with CREA-derived estimates using Pearson coefficients and intraclass correlation coefficients (ICC). The dual-sensor configuration consistently outperformed single-sensor models. For PB classification, the MH-CNN achieved the highest performance (F1-score 0.750). For EE, the RF model performed best (F1-score 0.741). Dual-sensor free-living estimates showed epidemiologically plausible distributions across the 24-hours period, including sitting 37% (538/1440 min), lying 34% (496/1440 min), walking 9% (131/1440 min) and moderate-to-vigorous intensity physical activity (MVPA) 2% (31/1440 min). Agreement with CREA was strongest for standing, walking, and cycling (r≥0.86; ICC≥0.72), while lying showed modest reliability (ICC=0.48). For EE, agreement was highest for light physical activity (LPA) and MVPA (ICC 0.72–0.75). Multitask models combining thigh- and waist-worn accelerometers, provide robust estimates of PB and EE under free-living conditions. The dual-sensor approach yielded more stable and epidemiologically coherent estimates than single-sensor methods, supporting its potential for large-scale population monitoring and mobile health applications. RR2-10.2196/preprints.70186

Background:Physical activity, sedentary behavior, sleep, and eating behavior are recognized as key contributors to physical and mental health. Ecological momentary assessment (EMA) can capture both the temporal variations and contextual correlates of these behaviors. Objective:This study aimed to identify, in adults, (1) multibehavioral clusters, including the social, physical, and psychological contexts as assessed by EMA; and (2) the associations of these behaviors with selected health-related outcomes. Methods:A sample of 510 participants from Czechia, Germany, France, and Ireland, with cross-sectional data collected in 2023-2024, was included (WEALTH study; median age 35.5, IQR 25.0-50.0 y; n=289, 56.7% female participants). During a 7-day free-living period, data on sedentary behavior (activPAL), physical activity (ActiGraph), sleep, eating behavior, and contextual characteristics (EMA) were collected. Dietary intake (Food Frequency Questionnaire) was assessed prior to the 7-day period. We used factor analysis of mixed data on 35 variables and then applied hierarchical clustering on principal components. Linear regression models with robust variance were used to examine associations between the clusters and health-related quality of life (36-Item Short Form Health Survey), well-being (5-item World Health Organization Well-Being Index), and handgrip strength. A multinomial logistic regression was used to examine the association between clusters and self-rated health. Results:Four multibehavioral clusters were identified: unhealthy behavior (n=109, reference group), mixed behavior-healthy diet (n=173), mixed behavior-stable good mood (n=142), and healthy behavior (n=86). The mixed behavior-stable good mood and healthy behavior clusters showed significantly higher average scores on mental health-related quality of life (β=6.32, 95% CI 3.73-8.91 and β=5.61, 95% CI 2.47-8.75, respectively) and well-being (β=11.62, 95% CI 7.86-15.38 and β=9.63, 95% CI 5.17-14.10, respectively) compared with the unhealthy behavior cluster. Participants in these clusters and those in the mixed behavior-healthy diet cluster tended to report better perceived health than those in the unhealthy behavior cluster (odds ratio [OR] 3.19, 95% CI 1.20-8.49; OR 9.85, 95% CI 3.54-27.40; OR 12.18, 95% CI 4.04-36.71, respectively). Conclusions:These findings indicate that engaging in multiple healthy lifestyle behaviors is associated with better health outcomes than focusing on a single domain and demonstrate the opportunity to identify meaningful multibehavioral patterns related to health by using EMA combined with body-worn movement sensors.

PURPOSE:The purpose of this study was to explore whether incorporating gyroscopic and accelerometer data will improve the prediction of energy expenditure (EE) of preschool children. Three model configurations were developed and compared using (1) accelerometer, (2) gyroscope, and (3) accelerometer + gyroscope data (dual sensor). METHOD:Participants (n = 39; aged 3 to <6 years) were equipped with OPAL, GT9X, and GENEActiv devices, worn on the right hip, right wrist, and left wrist, while EE was simultaneously measured using a portable metabolic unit. The protocol consisted of semistructured activities spanning a range of intensities from low to high. A total of 54 machine learning models were developed to predict EE (2 EE measures [metabolic equivalents, kilojoules per minute] × 3 wear locations × 3 model types [random forest, linear regression, and fully connected neural network] × 3 sensor configurations). Model performance was evaluated using root mean squared error. RESULTS:Our findings reveal that, across the various configurations, the random forest model utilizing dual-sensor data achieved marginally lower mean root mean squared error in the majority of cases. CONCLUSION:Given the minimal improvements observed and the challenges associated with data acquisition, we recommend that researchers utilize accelerometer-based models moving forward.

Background: This study explored the acceptability of an intensive ambulatory protocol combining four wearable devices and ecological momentary assessment (EMA) to assess physical and eating behaviors. Methods: EMA and wearable data were collected over 7 days in a convenience sample of 622 participants (56.1% women, Mage=38.2 years) in Ireland, Germany, France, and the Czech Republic. Short EMA questionnaires (eight to 17 items) assessing current activity, context, and mood were triggered randomly (seven per day) and in response to behavioral patterns detected by Fitbit, such as prolonged sitting (max four per day), walking, or running (each max three per day). EMA compliance was assessed by the proportion of completed questionnaires. Among four wearables worn (Fitbit/ActivPAL/Actigraph/LifeQ), compliance was measured using Fitbit heart rate recordings. Acceptability for EMA and wearables was assessed using 5-point Likert-scale questionnaires on daily burden, ease of use, interference, and ethical satisfaction. Associations with sociodemographic and health factors were analyzed. Results: Participants received on average 11.5 (+/- SD 1.9) EMA questionnaires per day. EMA acceptability scores ranged from 3.8 (burden) to 4.6 (ethical satisfaction), while wearable acceptability scores ranged from 4.1 (burden/reactivity) to 4.6 (ease of use/ ethical satisfaction). Low perceived daily burden was associated with higher compliance for EMA wearables (odds ratio: 1.88, 95% CI [1.30, 2.72] and 1.74, 95% CI [1.08, 2.80], respectively), while smartphone interference increased with age (odds ratio: 0.62, 95% CI [0.50, 0.78]). No association was found with sex and educational level. Conclusions: Acceptability of EMA and wearables was generally satisfactory. As EMA compliance may be affected by daily burden, well-designed protocols are key to leveraging ambulatory assessment of real-time behavioral and contextual data.

Abstract Environmental exposures are increasingly examined in relation to mental health, yet large-scale epidemiological analyses remain constrained by fragmented geospatial data, heterogeneous spatial and temporal resolutions, and privacy-preserving linkage requirements, limiting systematic investigation of multiple environmental domains at the population level. We present environMAP, a harmonised set of analysis-ready environmental exposure layers derived from open, global sources. environMAP spans the built environment, green and blue spaces, light exposure (solar radiation and night-time light), terrain, weather and extremes, and air pollution. We document data provenance, spatial buffers, preprocessing, projection alignment, and metadata, and provide a reproducible workflow for privacy-preserving linkage to cohort residential locations. To demonstrate utility, we linked environMAP to >200,000 adults in the German National Cohort (NAKO) and summarised self-reported lifetime doctor-diagnosed depression across exposure gradients using sex-stratified descriptive analyses. Gradients were interpretable and broadly consistent with prior evidence, supporting feasibility, scalability, and hypothesis generation. The framework is adaptable to other outcomes, cohorts, and regions.

Background:The accurate measurement of physical behaviors (PBs) and eating behaviors (EBs) is critical for designing, monitoring, and implementing public health guidelines and intervention strategies. The objective of the Wearable Sensor Assessment of Physical and Eating Behaviours (WEALTH) project was to develop standardized methods to identify daily PBs and EBs from wearable research- and consumer-grade sensors and evaluate the interaction and contexts of these behaviors. Objective:The aim of this paper is to describe the study design and methods and report on the descriptive characteristics of the participants. Methods:Within the framework of the WEALTH project, a cross-sectional study (spring 2023 to spring 2024) was completed in 5 European research centers in the Czech Republic, France, Germany, and Ireland. In each center, participants attended a research lab, completed an online questionnaire, and provided measures of anthropometry and handgrip strength. The participants were then fitted with 2 research-grade and 2 consumer-grade devices and participated in a standardized semistructured lab-based activity protocol. The latter was specifically designed to collect labeled data that simulated common PBs and EBs typical for a daily routine. Participants were then followed during a 9-day free-living data collection period, which combined the assessment of PB and EB via wearable devices and time-based, event-based, and self-initiated ecological momentary assessments (EMAs). The EMA surveys were complemented by three 24-hour dietary recalls, using validated web-based programs. Upon the completion of the survey protocol, participants completed a questionnaire that assessed the feasibility of the procedures. Results:The final sample includes 627 participants, of whom 44% (n=275) were male. The mean age was 32.7 (SD 13.3) years, and the mean body mass index was 24.5 (SD 4.0) kg/m². The WEALTH study data will be used to develop machine learning (ML) models for classifying daily activities from wrist and hip-worn accelerometer data, evaluate EMA methods for studying interactions between PB and EB, and evaluate the feasibility and compliance of the methods. Data processing and ML model development are currently underway, with primary results expected to be published in 2026. Conclusions:The output of the WEALTH project will be provided via a repository and a comprised toolbox of publicly available labeled data, ML models for behavior classification from accelerometer data, and a methodology to simultaneously capture EB and PB, thereby producing an integrated data collection system to support future research.

Early childhood is a period of rapid development; research shows that the formation of healthy habits during this period can result in higher physical fitness levels and better sleep, but also long-term improved mental health and wellbeing. Despite structures supporting physical activity (PA) and related behaviours, many German children under 6 years do not achieve the recommended levels of PA, sedentary behaviour (SED), and sleep; this in turn can hinder the formation healthy lifestyle habits in the early years, and also lead to long-term poor mental health, later in life. Thus, this study aimed to explore the associations between device-based measured PA with nighttime sleep, SED, and mental health in German children under six years of age. PA, sleep, and SED were assessed at baseline and 1-year follow-up using wrist-worn GENEActiv accelerometers sampled at 100 Hz. The R-package GGIR (version 3.1.1) was used to derive light-intensity PA (LPA), moderate-to-vigorous-intensity PA (MVPA), total PA (TPA), inactivity (proxy for SED), total night sleep time (TST), and sleep efficiency (SE). Parents answered on children’s mental health using the Strengths and Difficulties Questionnaire (SDQ). Linear mixed models, were used to estimate cross-sectional associations from repeated measures of PA intensities with sleep, SED, and mental health, adjusting for age and sex of the child, parental education, migration background, urbanity, and household income. We investigated 212 children aged 2–6 years (51

Objective. The first objective of this study was to refine previously designed machine learning models that predict energy expenditure (EE) of preschool children by modifying the method used to calculate metabolic equivalents (METs). The secondary objective was to compare estimates of time spent in different physical activity intensities across the newly developed models, previously published METs models, existing METs-based models from the literature, and models calibrated using direct observation.Approach. The model training dataset included 35 Canadian children (aged 3.0-5.99 years) equipped with GT9X accelerometers on their right hip. A portable metabolic unit was used to measure EE during a semi-structured protocol consisting of activities ranging from low- to high-intensity. The resulting models were applied to a sample of Canadian preschool children (n= 118; aged 3.0-5.99 years) to estimate time spent in sedentary (SED), light (LPA), moderate-to-vigorous (MVPA), and total physical activity (TPA). A repeated measures ANOVA was used to compare time estimates across models and according to three different configurations of METs activity thresholds.Main results. Results indicated that the newly developed models from Objective 1 produced significantly different estimates of time spent in SED, LPA, MVPA, and TPA compared to both previously published models and other existing METs-based models, highlighting the impact of different approaches to calculating METs.Significance. Model selection and METs calculation methods markedly influenced activity intensity estimates, underscoring the need for consistent methodology. Classification models yielded the most plausible free-living estimates.

External factors, including digital media (DM), promote body ideals that can shape adolescents’ body image, but studies across European countries are scarce. Therefore, the aim of the study was to examine the relationship between daily DM duration and body image dissatisfaction (BID) in adolescents from nine European countries. Participants from the I.Family study self-reported daily DM duration and BID in 2013/2014 (n = 3,608; 51

Background: Accurate classification of physical behavior from accelerometer data is crucial for health and behavioral research. While machine learning models often perform well within the populations they are trained on, they are rarely validated on independent populations, and their generalizability remains poorly understood. Therefore, we aimed to externally validate a widely used random forest model for physical behavior classification, and to assess whether its performance varied by participants' age, sex, or body mass index. Methods: We validated the random forest classifier, trained by Ellis et al., which achieved a balanced accuracy of 79% for classifying sitting, standing, and walking/running from hip-worn accelerometer data in the original training population. For the external validation, we obtained ActiGraph recordings for 610 participants from four European countries from the WEALTH (WEarable sensor Assessment of physicaL and eaTing beHaviors) project, which were labeled with the corresponding free-living behavior using ecological momentary assessment. Classifier performance was assessed using confusion matrices, precision, recall, Fscore, and balanced accuracy. Results: Inthe WEALTH population, the random forest classifier achieved a balanced accuracy of 40% and an average F-score of 0.33. Precision and recall were highest for sitting, followed by walking/running and standing. Performance was consistent across subpopulations defined by age, sex, and body mass index. Conclusion: The substantial reduction in accuracy demonstrates the limited generalizability of the existing random forest classifier. Our findings underscore the need for external validation and more diverse training data to ensure robust application of machine learning models in physical behavior research.

Observational studies have suggested that air pollution is associated with impaired glucose metabolism; yet randomized controlled trials to confirm the causality of this association are ethically and practically unfeasible. We emulated a hypothetical trial to evaluate the effects of sustained reductions in ambient air pollutants on homeostasis model assessment for insulin resistance (HOMA-IR) and fasting glucose (FG) in children and adolescents. Combining target trial emulation with g-computation, we estimated the effects of sustained hypothetical reductions of air pollutants on HOMA-IR and FG compared to no intervention (natural course). Our sample comprised 1417 children aged 2-9 years at baseline (2007/2008) participating in the pan-European IDEFICS/I.Family cohort. Ambient annual average levels of particulate matter<2.5 μm (PM2.5), black carbon (BC), and NO2 were estimated at residential addresses using land use regression models. We found a clear dose-response relationship between sustained reductions in PM2.5 and BC and decreasing levels of HOMA-IR and FG. Hypothetically reducing PM2.5 to WHO's recommended annual level of 5 μg/m3, the mean HOMA-IR z-score declined by -0.95 (95 % CI: -1.65; -0.44). Similarly, lowering BC to 0.8 × 10-5/m reduced HOMA-IR by -0.36 (95 % CI: -0.61; -0.12) and FG by -0.28 (95 % CI: -0.50; -0.04). We found no clear evidence of NO2 reductions on HOMA-IR and FG. Even at relatively low pollution levels in our cohort, further reductions in PM2.5 and BC can improve HOMA-IR and FG levels in children and adolescents. Our findings provide evidence for the development of BC recommendations for air quality guidelines.

Ecological Momentary Assessment (EMA) enables the real-time capture of health-related behaviours, their situational contexts, and associated subjective experiences. This study aimed to evaluate the feasibility of an EMA targeting physical and eating behaviours, optimise its protocol, and provide recommendations for future large-scale EMA data collections. The study involved 52 participants (age 31±9 years, 56% females) from Czechia, France, Germany, and Ireland completing a 9-day free-living EMA protocol using the HealthReact platform connected to a Fitbit tracker. The EMA protocol included time-based (7/day), event-based (up to 10/day), and self-initiated surveys, each containing 8 to 17 items assessing physical and eating behaviours and related contextual factors such as affective states, location, and company. Qualitative insights were gathered from post-EMA feedback interviews. Compliance was low (median 49%), particularly for event-based surveys (median 34%), and declined over time. Many participants were unable or unwilling to complete surveys in certain contexts (e.g., when with family), faced interference with their daily schedules, and encountered occasional technical issues, suggesting the need for thorough initial training, an individualised protocol, and systematic compliance monitoring. The number of event-based surveys was less than desired for the study, with a median of 2.4/day for sedentary events, when 4 were targeted, and 0.9/day for walking events, when 3 were targeted. Conducting simulations using participants' Fitbit data allowed for optimising the triggering rules, achieving the desired median number of sedentary and walking surveys (3.9/day for both) in similar populations. Self-initiated reports of meals and drinks yielded more reports than those prompted in time-based and event-based EMA surveys, suggesting that self-initiated surveys might better reflect actual eating behaviours. This study highlights the importance of assessing feasibility and optimising EMA protocols to enhance subsequent compliance and data quality. Conducting pre-tests to refine protocols and procedures, including simulations using participants' activity data for optimal event-based triggering rules, is crucial for successful large-scale data collection in EMA studies of physical and eating behaviours.

This study aimed to examine the impact of preprocessing and inclusion of various features on predicting the energy expenditure (EE) of preschool children (3.0–6.99 years). The internal Canadian sample consisted of 36 children, equipped with accelerometers on their wrists (OPAL) and right hip (ActiGraph GT9X). The external German sample consisted of 41 children, equipped with accelerometers on their wrists (GENEActiv) and right hip (GENEActiv; ActiGraph GT3X +). Both datasets used portable metabolic units to record EE. The effects of filtering, rectifying, adding a time delay, frequency domain (FD) features, and participant features on EE prediction across linear regression, random forest (RF), and fully connected neural network models. The Canadian sample was split into training (2/3) and validation (1/3) sets, and the German sample served as an external validation set. Consistently it was found that the RF with filtered, not rectified data with FD, participant features, and a time delay resulted in improved performance compared to approaches used previously. The models also performed similarly in the holdout sample but resulted in higher error when applied in the external validation dataset. Results attest that filtering, not rectifying, FD features and participant features result in improved model performance to predict the EE of preschool children.

Objective. This study aimed to develop convolutional neural networks (CNNs) models to predict the energy expenditure (EE) of children from raw accelerometer data. Additionally, this study sought to external validation of the CNN models in addition to the linear regression (LM), random forest (RF), and full connected neural network (FcNN) models published in Steenbock et al (2019 J. Meas. Phys. Behav. 2 94-102). Approach. Included in this study were 41 German children (3.0-6.99 years) for the training and internal validation who were equipped with GENEActiv, GT3X+, and activPAL accelerometers. The external validation dataset consisted of 39 Canadian children (3.0-5.99 years) that were equipped with OPAL, GT9X, GENEActiv, and GT3X+ accelerometers. EE was recorded simultaneously in both datasets using a portable metabolic unit. The protocols consisted of a semi-structured activities ranging from low to high intensities. The root mean square error (RMSE) values were calculated and used to evaluate model performances. Main results. (1) The CNNs outperformed the LM (13.17%-23.81% lower mean RMSE values), FcNN (8.13%-27.27% lower RMSE values) and the RF models (3.59%-18.84% lower RMSE values) in the internal dataset. (2) In contrast, it was found that when applied to the external Canadian dataset, the CNN models had consistently higher RMSE values compared to the LM, FcNN, and RF. Significance. Although CNNs can enhance EE prediction accuracy, their ability to generalize to new datasets and accelerometer brands/models, is more limited compared to LM, RF, and FcNN models.

Physical activity measured by accelerometry (PA-accelerometry) is used as an indicator of physical capacity in chronic diseases. Currently, only fragmented age ranges of reference percentile curves are available for European children and adolescents. This study aimed to provide age- and sex-specific percentiles for physical activity measured by hip-worn accelerometry derived throughout the full age range of European children and adolescents. Individual-level population-based PA data measured by accelerometry from HELENA and IDEFICS/I.Family studies were pooled and harmonized. Together these studies involved children and adolescents aged 2–18 years from 12 European countries. Primary outcomes included averaged counts per minute (CPM), sedentary time (SED), light PA (LPA) and moderate-to-vigorous PA (MVPA). Generalized Additive Models for Location, Scale and Shape were used to derive age- and sex-specific reference percentile curves for these outcomes. The combined cohort consisted of 11,645 children and adolescents aged 2 to 18 years who contributed 14,610 valid accelerometry recordings, with a median accelerometer wear time of 6 days. This dataset allowed for the construction of age- and sex-specific reference percentile curves for CPM, SED, LPA, and MVPA. The curves demonstrated varying trends and variability across age groups. Conclusions: This study provides age- and sex-specific percentile curves for PA-accelerometry in European children and adolescents, addressing a current gap in the availability of full-age range reference data. These curves based on healthy children and adolescents can be used by clinicians, researchers, and policymakers to interpret PA-accelerometry measurements, track physical activity trends, and evaluate treatment responses and health interventions.

The accurate measurement of physical and eating behaviours is critical for designing, monitoring, and implementing public health guidelines and intervention strategies. The objective of the Wearable Sensor Assessment of Physical and Eating Behaviours (WEALTH) project was to develop standardised methods to identify daily physical and eating behaviours from wearable research- and consumer-grade sensors and to evaluate the interaction and contexts of these behaviours. The aim of this paper is to describe the study design and methods, and report on the descriptive characteristics of the participants. Within the framework of the WEALTH project, a cross-sectional study (spring 2023 to spring 2024) was completed in five European research centres in the Czech Republic, France, Germany, and Ireland. In each centre, participants attended the research lab, completed an online questionnaire and provided measures of anthropometry and handgrip strength. The participants were then fitted with two research-grade (ActiGraph wGT3X-BT, activPAL 3 micro) and two consumer-grade (Fitbit® Charge 5, LifeQ® enabled smartwatch) devices and participated in a standardised semi-structured lab-based activity protocol. The latter was specifically designed to collect labelled data that simulated common physical behaviours (PB) and eating behaviours (EB) typical for a daily routine. Participants were then followed during a 9-day free-living data collection period which combined the assessment of PB and EB via wearable devices and time-based, event-based and self-initiated ecological momentary assessments (EMA). The EMA surveys were complemented by three 24-hour dietary recalls, using validated web-based programs. Upon completion of the survey protocol, participants completed a questionnaire that assessed the feasibility of the procedures. The WEALTH study data will be used to develop machine learning (ML) models for classifying daily activities from wrist and hip worn accelerometer data, to evaluate EMA methods for studying interactions between PB and EB and to evaluate feasibility and compliance of the methods. Further analyses will provide insights in the association of classified PB and related EB examining behavioural patterns and health outcomes.The final sample was 627 participants, of which 44% were male. The mean age was 32.7 years (± 13.3), and the mean body mass index was 24.5 kg/m² (± 4.0). The output of the WEALTH project will be provided via a repository and a comprised toolbox of publicly available labelled data, ML models for behaviour classification from accelerometer data and a methodology to simultaneously capture EB and PB, thereby producing an integrated data collection system to support future research.

Large population-based cohort studies utilizing device-based measures of physical activity are crucial to close important research gaps regarding the potential protective effects of physical activity on chronic diseases. The present study details the quality control processes and the derivation of physical activity metrics from 100 Hz accelerometer data collected in the German National Cohort (NAKO). During the 2014 to 2019 baseline assessment, a subsample of NAKO participants wore a triaxial ActiGraph accelerometer on their right hip for seven consecutive days. Auto-calibration, signal feature calculations including Euclidean Norm Minus One (ENMO) and Mean Amplitude Deviation (MAD), identification of non-wear time, and imputation, were conducted using the R package GGIR version 2.10-3. A total of 73,334 participants contributed data for accelerometry analysis, of whom 63,236 provided valid data. The average ENMO was 11.7 ± 3.7 mg (milli gravitational acceleration) and the average MAD was 19.9 ± 6.1 mg. Notably, acceleration summary metrics were higher in men than women and diminished with increasing age. Work generated in the present study will facilitate harmonized analysis, reproducibility, and utilization of NAKO accelerometry data. The NAKO accelerometry dataset represents a valuable asset for physical activity research and will be accessible through a specified application process.