Activity monitoring combined with machine learning (ML) methods can contribute to detailed knowledge about daily physical behavior in older adults. The current study (1) evaluated the performance of an existing activity type recognition ML model (HARTH), based on data from healthy young adults, for classifying daily physical behavior in fit-to-frail older adults, (2) compared the performance with a ML model (HAR70+) that included training data from older adults, and (3) evaluated the ML models on older adults with and without walking aids. Eighteen older adults aged 70–95 years who ranged widely in physical function, including usage of walking aids, were equipped with a chest-mounted camera and two accelerometers during a semi-structured free-living protocol. Labeled accelerometer data from video analysis was used as ground truth for the classification of walking, standing, sitting, and lying identified by the ML models. Overall accuracy was high for both the HARTH model (91%) and the HAR70+ model (94%). The performance was lower for those using walking aids in both models, however, the overall accuracy improved from 87% to 93% in the HAR70+ model. The validated HAR70+ model contributes to more accurate classification of daily physical behavior in older adults that is essential for future research.

Annotating accelerometer-based physical activity data remains a challenging task, limiting the creation of robust supervised machine learning models due to the scarcity of large, labeled, free-living human activity recognition (HAR) datasets. Researchers are exploring self-supervised learning (SSL) as an alternative to relying solely on labeled data approaches. However, there has been limited exploration of the impact of large-scale, unlabeled datasets for SSL pre-training on downstream HAR performance, particularly utilizing more than one accelerometer. To address this gap, a transformer encoder network is pre-trained on various amounts of unlabeled, dual-accelerometer data from the HUNT4 dataset: 10, 100, 1k, 10k, and 100k hours. The objective is to reconstruct masked segments of signal spectrograms. This pre-trained model, termed SelfPAB, serves as a feature extractor for downstream supervised HAR training across five datasets (HARTH, HAR70+, PAMAP2, Opportunity, and RealWorld). SelfPAB outperforms purely supervised baselines and other SSL methods, demonstrating notable enhancements, especially for activities with limited training data. Results show that more pre-training data improves downstream HAR performance, with the 100k-hour model exhibiting the highest performance. It surpasses purely supervised baselines by absolute F1-score improvements of 7.1% (HARTH), 14% (HAR70+), and an average of 11.26% across the PAMAP2, Opportunity, and RealWorld datasets. Compared to related SSL methods, SelfPAB displays absolute F1-score enhancements of 10.4% (HARTH), 18.8% (HAR70+), and 16% (average across PAMAP2, Opportunity, RealWorld).

Background: Physical activity (PA) is imperative for healthy ageing and is a modifiable lifestyle factor. Accurate, clinically meaningful estimates of daily PA among older adults can inform targeted interventions to maintain function and independence. This study describes daily PA in older adults across levels of care as a first step contributing to the limited evidence on potential associations between PA and the use of care services. Methods: This study used data from the Trondheim 70 + cohort in the population-based Norwegian HUNT Study. In total, 1042 participants aged 70 years or older with valid activity data were included. PA was assessed using two accelerometers over 7 consecutive days and was classified into PA (walking, standing, running, and cycling) and sedentary behavior (sitting and lying). Data on received care services were retrieved from municipal registers and participants were classified into four levels of care: 1) independently living (81.9%), 2) independently living with low-level home care services (6.5%), 3) recipients of home care services (6.0%), and 4) nursing home residents (5.7%). Time spent in the activity types and duration of bouts are presented across levels of care. Results: Participants mean age was 77.5 years (range: 70.1–105.4, 55% female) and PA was lower with higher age. Across levels of care, significant group differences were found in the total time spent in PA, particularly in walking and standing. Daily PA, duration of active bouts, and number of daily walking bouts were lower for participants receiving higher levels of care. Standing was the dominant type of PA and walking appeared predominantly in short bouts at all care levels. Conclusions: This is the first population-based study using device-measured PA to describe daily PA across levels of care. The results showed that low-intensity activities constitute the primary component of everyday PA, advocating for placing greater emphasis on the significant role these activities play in maintaining daily PA at older age. Furthermore, the study demonstrated that activity types and bout durations are related to the ability to live independently among older adults. Overall, these findings can contribute to better target interventions to maintain function and independence in older adults.

Human Activity Recognition models have potential to contribute to valuable and detailed knowledge of habitual physical activity for typically developing children and children with Cerebral Palsy. The main objective of the present study was to develop and validate two Human Activity Recognition models. One trained on data from typically developing children (n = 63), the second also including data from children with Cerebral Palsy (n = 16), engaging in standardised activities and free play. Our data was collected using accelerometers and ground truth was established with video annotations. Additionally, we aimed to investigate the influence of window settings on model performance. Utilizing the Extreme gradient boost (XGBoost) classifier, twelve sub-models were created, with 1-,3- and 5-seconds windows, with and without overlap. Both Human Activity Recognition models demonstrated excellent predictive capabilities (>92%) for standardised activities for both typically developing and Cerebral Palsy. From all window sizes, the 1-second window performed best for all test groups. Accuracy was slightly lower (>75%) for the Cerebral Palsy test group performing free play activities. The impact of window size and overlap varied depending on activity. In summary both Human Activity Recognition models effectively predict standardised activities, surpassing prior models for typically developing and children with Cerebral Palsy. Notably, the model trained on combined typically developing children and Cerebral Palsy data performed exemplary across all test groups. Researchers should select window settings aligned with their specific research objectives.

Background: Impaired postural control is a key feature of cerebral palsy that affects daily living. Measures of trunk movement and acceleration have been used to assess dynamic postural control previously. In many children with cerebral palsy, ankle-foot orthoses are used to provide a stable base of support, but their effect on postural control is not yet understood. Objectives: The objectives of the current study were to investigate the effects of ankle-foot orthoses on postural control and energy cost of walking in children with cerebral palsy. Study design: Clinical study with controls. Methods: Trunk accelerometry (amplitude and structure) and energy cost of walking (J/kg/m) were recorded from five-minute walking trials with and without ankle-foot orthoses for children with cerebral palsy and without orthoses for the reference group. Results: Nineteen children with unilateral spastic cerebral palsy and fourteen typically developed children participated. The use of ankle-foot orthoses increased structure complexity of trunk acceleration in mediolateral and anterior–posterior directions. The use of ankle-foot orthoses changed mediolateral-structure toward values found in typically developed children. This change was not associated with a change in energy cost during walking. Conclusions: The use of ankle-foot orthoses does affect trunk acceleration that may indicate a beneficial effect on postural control. Using measures of trunk acceleration may contribute to clinical understanding on how the use of orthoses affect postural control.

Introduction: Understanding factors that influence care service use is crucial for developing preventive strategies to maintain independence among older adults. In this study, we aimed to identify distinct trajectory groups of municipal care service use among community-dwelling older adults to determine whether daily physical activity is associated with future care service use.Methods: This prospective cohort study included 981 community-dwelling older adults from the HUNT4 Trondheim 70+ study. At baseline, physical activity was assessed over seven consecutive days using two accelerometers attached to the thigh and lower back. An activity type machine learning model was used to classify the physical activity types: walking, standing, cycling, running, sitting, and lying. Municipal care service use was retrieved monthly from medical records for 3 years. Using group-based trajectory modeling, we identified distinct trajectories of care service use. Multinomial regression models adjusted for age, sex, education level, dementia, and physical performance were used to evaluate the associations between daily physical activity at baseline and care service group belonging.Results: We identified four distinct trajectory groups of municipal care service use, labeled steady low (72.7%), low increasing (9.0%), medium increasing (12.0%), and high increasing (6.3%). Daily time spent in total physical activity was not associated with trajectory group belonging when adjusted for age, sex, education level, dementia, and physical performance. However, more time spent walking, in bouts lasting longer than a minute, was associated with a reduced relative risk of belonging to the high increasing compared to the steady low group. Furthermore, age, physical performance, and dementia were all significantly associated with trajectory group belonging, and sex differences were observed. Compared to women, men had a reduced relative risk of belonging to the low increasing, medium increasing, or high increasing trajectory groups.Conclusion: This study identified four distinct trajectories of municipal care service use among older adults over 3 years. Total daily physical activity was not associated with trajectories of care service use, but more time spent walking in longer bouts was independently associated with lower care service use, even when adjusted for the strong predictors of physical performance, dementia diagnosis, and age.