Segmenting accelerometer data from daily life with unsupervised machine learning

A novel ViT-BILSTM model for physical activity intensity classification in adults using gravity-based acceleration

AIM

The aim of this study is to apply a novel hybrid framework incorporating a Vision Transformer (ViT) and bidirectional long short-term memory (Bi-LSTM) model for classifying physical activity intensity (PAI) in adults using gravity-based acceleration. Additionally, it further investigates how PAI and temporal window (TW) impacts the model' s accuracy.

METHOD

This research used the Capture-24 dataset, consisting of raw accelerometer data from 151 participants aged 18 to 91. Gravity-based acceleration was utilised to generate images encoding various PAIs. These images were subsequently analysed using the ViT-BiLSTM model, with results presented in confusion matrices and compared with baseline models. The model's robustness was evaluated through temporal stability testing and examination of accuracy and loss curves.

RESULT

The ViT-BiLSTM model excelled in PAI classification task, achieving an overall accuracy of 98.5% ± 1.48% across five TWs-98.7% for 1s, 98.1% for 5s, 98.2% for 10s, 99% for 15s, and 98.65% for 30s of TW. The model consistently exhibited superior accuracy in predicting sedentary (98.9% ± 1%) compared to light physical activity (98.2% ± 2%) and moderate-to-vigorous physical activity (98.2% ± 3%). ANOVA showed no significant accuracy variation across PAIs (F = 2.18, p = 0.13) and TW (F = 0.52, p = 0.72). Accuracy and loss curves show the model consistently improves its performance across epochs, demonstrating its excellent robustness.

CONCLUSION

This study demonstrates the ViT-BiLSTM model's efficacy in classifying PAI using gravity-based acceleration, with performance remaining consistent across diverse TWs and intensities. However, PAI and TW could result in slight variations in the model's performance. Future research should concern and investigate the impact of gravity-based acceleration on PAI thresholds, which may influence model's robustness and reliability.

Physical activity in young children across developmental and health states: the ActiveCHILD study

Background

Evidence about physical activity of young children across developmental and health states is very limited. Using data from an inclusive UK cohort, ActiveCHILD, we investigated relationships between objectively measured physical activity, child development, social context, and health-related quality of life (HRQoL).

Methods

Children (12-36 months), purposively sampled across health pathways, developmental abilities, and sociodemographic factors, were recruited through thirteen National Health Service organisations in England. Data were collected from 07/2017 to 08/2019 on: weekly physical activity (3-7 days) using waist-worn accelerometer (ActiGraph 3GTX); sociodemographics, parent actions, child HRQoL, and child development using questionnaires; and child health conditions using clinical records. A data-driven, unsupervised method, called hidden semi-Markov model (HSMM) segmented the accelerometery data and provided estimates of the total time spent active (any intensity) and very active (greater intensity) for each child. Relationships with the explanatory factors were investigated using multiple linear regression.

Findings

Physical activity data were obtained for 282 children (56% females, mean age 21 months, 37.5% with a health condition) covering all index of multiple deprivation deciles. The patterns of physical activity consisted of two daily peaks, children spending 6.44 (SD = 1.39) hours active (any intensity), of which 2.78 (SD = 1.38) hours very active, 91% meeting WHO guidelines. The model for total time active (any intensity) explained 24% of variance, with mobility capacity the strongest predictor (β = 0.41). The model for time spent very active explained 59% of variance, with mobility capacity again the strongest predictor (β = 0.76). There was no evidence of physical activity explaining HRQoL.

Interpretation

The findings provide new evidence that young children across developmental states regularly achieve mainstream recommended physical activity levels and challenges the belief that children with development problems need lower expectations for daily physical activity compared to peers. Advancing the rights of all children to participate in physical activity requires inclusive, equally ambitious, expectations for all.

Funding

Niina Kolehmainen, HEE/NIHR Integrated Clinical Academic Senior Clinical Lecturer, NIHR ICA-SCL-2015-01-00, was funded by the NIHR for this research project. Christopher Thornton, Olivia Craw, Laura Kudlek, and Laura Cutler were also funded from this award. Tim Rapley is a member of the NIHR Applied Research Collaboration North East and North Cumbria, with part of his time funded through the related award (NIHR200173). The views expressed in this publication are those of the author(s) and not necessarily those of the NIHR, NHS, or the UK Department of Health and Social Care. The work of Kianoush Nazarpour is supported by Engineering and Physical Sciences Research Council (EPSRC), under grant number EP/R004242/2.

Relationship between physical activity and central sensitization in chronic low back pain: Insights from machine learning

BACKGROUND AND OBJECTIVES

Chronic low back pain (CLBP) is a leading cause of disability. The management guidelines for the management of CLBP often recommend optimizing physical activity (PA). Among a subsample of patients with CLBP, central sensitization (CS) is present. However, knowledge about the association between PA intensity patterns, CLBP, and CS is limited. The objective PA computed by conventional approaches (e.g. cut-points) may not be sensitive enough to explore this association. This study aimed to investigate PA intensity patterns in patients with CLBP and low or high CS (CLBP-, CLBP+, respectively) by using advanced unsupervised machine learning approach, Hidden semi-Markov model (HSMM).

METHODS

Forty-two patients were included (23 CLBP-, 19 CLBP+). CS-related symptoms (e.g. fatigue, sensitivity to light, psychological features) were assessed by a CS Inventory. Patients wore a standard 3D-accelerometer for one week and PA was recorded. The conventional cut-points approach was used to compute the time accumulation and distribution of PA intensity levels in a day. For the two groups, two HSMMs were developed to measure the temporal organization of and transition between hidden states (PA intensity levels), based on the accelerometer vector magnitude.

RESULTS

Based on the conventional cut-points approach, no significant differences were found between CLBP- and CLBP+ groups (p = 0.87). In contrast, HSMMs revealed significant differences between the two groups. For the 5 identified hidden states (rest, sedentary, light PA, light locomotion, and moderate-vigorous PA), the CLBP- group had a higher transition probability from rest, light PA, and moderate-vigorous PA states to the sedentary state (p < 0.001). In addition, the CBLP- group had a significantly shorter bout duration of the sedentary state (p < 0.001). The CLBP+ group exhibited longer durations of active (p < 0.001) and inactive states (p = 0.037) and had higher transition probabilities between active states (p < 0.001).

CONCLUSIONS

HSMM discloses the temporal organization and transitions of PA intensity levels based on accelerometer data, yielding valuable and detailed clinical information. The results imply that patients with CLBP- and CLBP+ have different PA intensity patterns. CLBP+ patients may adopt the distress-endurance response pattern with a prolonged bout duration of activity engagement.

Using unsupervised machine learning to quantify physical activity from accelerometry in a diverse and rapidly changing population

Accelerometers are widely used to measure physical activity behaviour, including in children. The traditional method for processing acceleration data uses cut points to define physical activity intensity, relying on calibration studies that relate the magnitude of acceleration to energy expenditure. However, these relationships do not generalise across diverse populations and hence they must be parametrised for each subpopulation (e.g., age groups) which is costly and makes studies across diverse populations and over time difficult. A data-driven approach that allows physical activity intensity states to emerge from the data, without relying on parameters derived from external populations, offers a new perspective on this problem and potentially improved results. We applied an unsupervised machine learning approach, namely a hidden semi-Markov model, to segment and cluster the raw accelerometer data recorded (using a waist-worn ActiGraph GT3X+) from 279 children (9-38 months old) with a diverse range of developmental abilities (measured using the Paediatric Evaluation of Disability Inventory-Computer Adaptive Testing measure). We benchmarked this analysis with the cut points approach, calculated using thresholds from the literature which had been validated using the same device and for a population which most closely matched ours. Time spent active as measured by this unsupervised approach correlated more strongly with PEDI-CAT measures of the child's mobility (R2: 0.51 vs 0.39), social-cognitive capacity (R2: 0.32 vs 0.20), responsibility (R2: 0.21 vs 0.13), daily activity (R2: 0.35 vs 0.24), and age (R2: 0.15 vs 0.1) than that measured using the cut points approach. Unsupervised machine learning offers the potential to provide a more sensitive, appropriate, and cost-effective approach to quantifying physical activity behaviour in diverse populations, compared to the current cut points approach. This, in turn, supports research that is more inclusive of diverse or rapidly changing populations.

Examination of physical activity development in early childhood: protocol for a longitudinal cohort study of mother-toddler dyads

BACKGROUND

Physical activity (PA) development in toddlers (age 1 and 2 years) is not well understood, partly because of a lack of analytic tools for accelerometer-based data processing that can accurately evaluate PA among toddlers. This has led to a knowledge gap regarding how parenting practices around PA, mothers' PA level, mothers' parenting stress, and child developmental and behavioral problems influence PA development in early childhood.

METHODS

The Child and Mother Physical Activity Study is a longitudinal study to observe PA development in toddlerhood and examine the influence of personal and parental characteristics on PA development. The study is designed to refine and validate an accelerometer-based machine learning algorithm for toddler activity recognition (Aim 1), apply the algorithm to compare the trajectories of toddler PA levels in males and females age 1-3 years (Aim 2), and explore the association between gross motor development and PA development in toddlerhood, as well as how parenting practices around PA, mothers' PA, mothers' parenting stress, and child developmental and behavioral problems are associated with toddlerhood PA development (Exploratory Aims 3a-c).

DISCUSSION

This study will be one of the first to use longitudinal data to validate a machine learning activity recognition algorithm and apply the algorithm to quantify free-living ambulatory movement in toddlers. The study findings will help fill a significant methodological gap in toddler PA measurement and expand the body of knowledge on the factors influencing early childhood PA development.

The Dilemma of Analyzing Physical Activity and Sedentary Behavior with Wrist Accelerometer Data: Challenges and Opportunities

Physical behaviors (e.g., physical activity and sedentary behavior) have been the focus among many researchers in the biomedical and behavioral science fields. The recent shift from hip- to wrist-worn accelerometers in these fields has signaled the need to develop novel approaches to process raw acceleration data of physical activity and sedentary behavior. However, there is currently no consensus regarding the best practices for analyzing wrist-worn accelerometer data to accurately predict individuals' energy expenditure and the times spent in different intensities of free-living physical activity and sedentary behavior. To this end, accurately analyzing and interpreting wrist-worn accelerometer data has become a major challenge facing many clinicians and researchers. In response, this paper attempts to review different methodologies for analyzing wrist-worn accelerometer data and offer cutting edge, yet appropriate analysis plans for wrist-worn accelerometer data in the assessment of physical behavior. In this paper, we first discuss the fundamentals of wrist-worn accelerometer data, followed by various methods of processing these data (e.g., cut points, steps per minute, machine learning), and then we discuss the opportunities, challenges, and directions for future studies in this area of inquiry. This is the most comprehensive review paper to date regarding the analysis and interpretation of free-living physical activity data derived from wrist-worn accelerometers, aiming to help establish a blueprint for processing wrist-derived accelerometer data.

Timing of objectively-collected physical activity in relation to body weight and metabolic health in sedentary older people: a cross-sectional and prospective analysis

BACKGROUND

Little is known about the impact of timing as opposed to frequency and intensity of daily physical activity on metabolic health. Therefore, we assessed the association between accelerometery-based daily timing of physical activity and measures of metabolic health in sedentary older people.

METHODS

Hourly mean physical activity derived from wrist-worn accelerometers over a 6-day period was collected at baseline and after 3 months in sedentary participants from the Active and Healthy Ageing study. A principal component analysis (PCA) was performed to reduce the number of dimensions (e.g. define periods instead of separate hours) of hourly physical activity at baseline and change during follow-up. Cross-sectionally, a multivariable-adjusted linear regression analysis was used to associate the principal components, particularly correlated with increased physical activity in data-driven periods during the day, with body mass index (BMI), fasting glucose and insulin, HbA1c and the homeostatic model assessment for insulin resistance (HOMA-IR). For the longitudinal analyses, we calculated the hourly changes in physical activity and change in metabolic health after follow-up.

RESULTS

We included 207 individuals (61.4% male, mean age: 64.8 [SD 2.9], mean BMI: 28.9 [4.7]). Higher physical activity in the early morning was associated with lower fasting glucose (-2.22%, 95% CI: -4.19, -0.40), fasting insulin (-13.54%, 95%CI: -23.49, -4.39), and HOMA-IR (-16.07%, 95%CI: -27.63, -5.65). Higher physical activity in the late afternoon to evening was associated with lower BMI (-2.84%, 95% CI: -4.92, -0.70). Higher physical activity at night was associated with higher BMI (2.86%, 95% CI: 0.90, 4.78), fasting glucose (2.57%, 95% CI: 0.70, 4.30), and HbA1c (2.37%, 95% CI: 1.00, 3.82). Similar results were present in the prospective analysis.

CONCLUSION

Specific physical activity timing patterns were associated with more beneficial metabolic health, suggesting particular time-dependent physical activity interventions might maximise health benefits.

Feature selection for unsupervised machine learning of accelerometer data physical activity clusters - A systematic review

BACKGROUND

Identifying clusters of physical activity (PA) from accelerometer data is important to identify levels of sedentary behaviour and physical activity associated with risks of serious health conditions and time spent engaging in healthy PA. Unsupervised machine learning models can capture PA in everyday free-living activity without the need for labelled data. However, there is scant research addressing the selection of features from accelerometer data. The aim of this systematic review is to summarise feature selection techniques applied in studies concerned with unsupervised machine learning of accelerometer-based device obtained physical activity, and to identify commonly used features identified through these techniques. Feature selection methods can reduce the complexity and computational burden of these models by removing less important features and assist in understanding the relative importance of feature sets and individual features in clustering.

METHOD

We conducted a systematic search of Pubmed, Medline, Google Scholar, Scopus, Arxiv and Web of Science databases to identify studies published before January 2021 which used feature selection methods to derive PA clusters using unsupervised machine learning models.

RESULTS

A total of 13 studies were eligible for inclusion within the review. The most popular feature selection techniques were Principal Component Analysis (PCA) and correlation-based methods, with k-means frequently used in clustering accelerometer data. Cluster quality evaluation methods were diverse, including both external (e.g. cluster purity) or internal evaluation measures (silhouette score most frequently). Only four of the 13 studies had more than 25 participants and only four studies included two or more datasets.

CONCLUSION

There is a need to assess multiple feature selection methods upon large cohort data consisting of multiple (3 or more) PA datasets. The cut-off criteria e.g. number of components, pairwise correlation value, explained variance ratio for PCA, etc. should be expressly stated along with any hyperparameters used in clustering.

Novel approach to modeling high-frequency activity data to assess therapeutic effects of analgesics in chronic pain conditions

Osteoarthritis (OA) is a chronic condition often associated with pain, affecting approximately fourteen percent of the population, and increasing in prevalence. A globally aging population have made treating OA-associated pain as well as maintaining mobility and activity a public health priority. OA affects all mammals, and the use of spontaneous animal models is one promising approach for improving translational pain research and the development of effective treatment strategies. Accelerometers are a common tool for collecting high-frequency activity data on animals to study the effects of treatment on pain related activity patterns. There has recently been increasing interest in their use to understand treatment effects in human pain conditions. However, activity patterns vary widely across subjects; furthermore, the effects of treatment may manifest in higher or lower activity counts or in subtler ways like changes in the frequency of certain types of activities. We use a zero inflated Poisson hidden semi-Markov model to characterize activity patterns and subsequently derive estimators of the treatment effect in terms of changes in activity levels or frequency of activity type. We demonstrate the application of our model, and its advance over traditional analysis methods, using data from a naturally occurring feline OA-associated pain model.

A Novel Combination of Accelerometry and Ecological Momentary Assessment for Post-Stroke Paretic Arm/Hand Use: Feasibility and Validity

Use of the paretic arm and hand is a key indicator of recovery and reintegration after stroke. A sound methodology is essential to comprehensively identify the possible factors impacting daily arm/hand use behavior. We combined ecological momentary assessment (EMA), a prompt methodology capturing real-time psycho-contextual factors, with accelerometry to investigate arm/hand behavior in the natural environment. Our aims were to determine (1) feasibility and (2) measurement validity of the combined methodology. We monitored 30 right-dominant, mild-moderately motor impaired chronic stroke survivors over 5 days (6 EMA prompts/day with accelerometers on each wrist). We observed high adherence for accelerometer wearing time (80.3%), EMA prompt response (84.6%), and generally positive user feedback upon exit interview. The customized prompt schedule and the self-triggered prompt option may have improved adherence. There was no evidence of EMA response bias nor immediate measurement reactivity. An unexpected small but significant increase in paretic arm/hand use was observed over days (12–14 min), which may be the accumulated effect of prompting that provided a reminder to choose the paretic limb. Further research that uses this combined methodology is needed to develop targeted interventions that effectively change behavior and enable reintegration post-stroke.

Advanced analytical methods to assess physical activity behaviour using accelerometer raw time series data: a protocol for a scoping review

BACKGROUND

Physical activity (PA) is a complex multidimensional human behaviour. Currently, there is no standardised approach for measuring PA using wearable accelerometers in health research. The total volume of PA is an important variable because it includes the frequency, intensity and duration of activity bouts, but it reduces them down to a single summary variable. Therefore, analytical approaches using accelerometer raw time series data taking into account the way PA are accumulated over time may provide more clinically relevant features of physical behaviour. Advances on these fields are highly needed in the context of the rapid development of digital health studies using connected trackers and smartwatches. The objective of this review will be to map advanced analytical approaches and their multidimensional summary variables used to provide a comprehensive picture of PA behaviour.

METHODS

This scoping review will be guided by the Arksey and O'Malley methodological framework. A search for relevant publications will be undertaken in MEDLINE (PubMed), Embase and Web of Science databases. The selection of articles will be limited to studies published in English from January 2010 onwards. Studies including analytical methods that go beyond total PA volume, average daily acceleration and the conventional cut-point approaches, involving tri-axial accelerometer data will be included. Two reviewers will independently screen all citations, full-text articles and extract data. The data will be collated, stored and charted to provide a descriptive summary of the analytical methods and outputs, their strengths and limitations and their association with different health outcomes.

DISCUSSION

This protocol describes a systematic method to identify, map and synthesise advanced analytical approaches and their multidimensional summary variables used to investigate PA behaviour and identify potentially clinically relevant features. The results of this review will be useful to guide future research related to analysing PA patterns, investigate their association with health conditions and suggest appropriate recommendations for changes in PA behaviour. The results may be of interest to sports scientists, clinical researchers, epidemiologists and smartphone application developers in the field of PA assessment.

SCOPING REVIEW REGISTRATION

This protocol has been registered with the Open Science Framework (OSF): https://osf.io/yxgmb .

Calibrating Wrist-Worn Accelerometers for Physical Activity Assessment in Preschoolers: Machine Learning Approaches

BACKGROUND

Physical activity (PA) level is associated with multiple health benefits during early childhood. However, inconsistency in the methods for quantification of PA levels among preschoolers remains a problem.

OBJECTIVE

This study aimed to develop PA intensity cut points for wrist-worn accelerometers by using machine learning (ML) approaches to assess PA in preschoolers.

METHODS

Wrist- and hip-derived acceleration data were collected simultaneously from 34 preschoolers on 3 consecutive preschool days. Two supervised ML models, receiver operating characteristic curve (ROC) and ordinal logistic regression (OLR), and one unsupervised ML model, k-means cluster analysis, were applied to establish wrist-worn accelerometer vector magnitude (VM) cut points to classify accelerometer counts into sedentary behavior, light PA (LPA), moderate PA (MPA), and vigorous PA (VPA). Physical activity intensity levels identified by hip-worn accelerometer VM cut points were used as reference to train the supervised ML models. Vector magnitude counts were classified by intensity based on three newly established wrist methods and the hip reference to examine classification accuracy. Daily estimates of PA were compared to the hip-reference criterion.

RESULTS

In total, 3600 epochs with matched hip- and wrist-worn accelerometer VM counts were analyzed. All ML approaches performed differently on developing PA intensity cut points for wrist-worn accelerometers. Among the three ML models, k-means cluster analysis derived the following cut points: ≤2556 counts per minute (cpm) for sedentary behavior, 2557-7064 cpm for LPA, 7065-14532 cpm for MPA, and ≥14533 cpm for VPA; in addition, k-means cluster analysis had the highest classification accuracy, with more than 70% of the total epochs being classified into the correct PA categories, as examined by the hip reference. Additionally, k-means cut points exhibited the most accurate estimates on sedentary behavior, LPA, and VPA as the hip reference. None of the three wrist methods were able to accurately assess MPA.

CONCLUSIONS

This study demonstrates the potential of ML approaches in establishing cut points for wrist-worn accelerometers to assess PA in preschoolers. However, the findings from this study warrant additional validation studies.

Physical activity assessment by accelerometry in people with heart failure

Background

International guidelines for physical activity recommend at least 150 min per week of moderate-to-vigorous physical activity (MVPA) for adults, including those with cardiac disease. There is yet to be consensus on the most appropriate way to categorise raw accelerometer data into behaviourally relevant metrics such as intensity, especially in chronic disease populations. Therefore the aim of this study was to estimate acceleration values corresponding to inactivity and MVPA during daily living activities of patients with heart failure (HF), via calibration with oxygen consumption (VO₂) and to compare these values to previously published, commonly applied PA intensity thresholds which are based on healthy adults.

Methods

Twenty-two adults with HF (mean age 71 ± 14 years) undertook a range of daily living activities (including laying down, sitting, standing and walking) whilst measuring PA via wrist- and hip-worn accelerometers and VO₂ via indirect calorimetry. Raw accelerometer output was used to compute PA in units of milligravity (mg). Energy expenditure across each of the activities was converted into measured METs (VO₂/resting metabolic rate) and standard METs (VO₂/3.5 ml/kg/min). PA energy costs were also compared with predicted METs in the compendium of physical activities. Location specific activity intensity thresholds were established via multilevel mixed effects linear regression and receiver operator characteristic curve analysis. A leave-one-out method was used to cross-validate the thresholds.

Results

Accelerometer values corresponding with intensity thresholds for inactivity (< 1.5METs) and MVPA (≥3.0METs) were > 50% lower than previously published intensity thresholds for both wrists and waist accelerometers (inactivity: 16.7 to 18.6 mg versus 45.8 mg; MVPA: 43.1 to 49.0 mg versus 93.2 to 100 mg). Measured METs were higher than both standard METs (34–35%) and predicted METs (45–105%) across all standing and walking activities.

Conclusion

HF specific accelerometer intensity thresholds for inactivity and MVPA are lower than previously published thresholds based on healthy adults, due to lower resting metabolic rate and greater energy expenditure during daily living activities for HF patients.

Trial registration

Clinical trials.gov NCT03659877, retrospectively registered on September 6th 2018.

Evaluating the performance of raw and epoch non-wear algorithms using multiple accelerometers and electrocardiogram recordings

Accurate detection of accelerometer non-wear time is crucial for calculating physical activity summary statistics. In this study, we evaluated three epoch-based non-wear algorithms (Hecht, Troiano, and Choi) and one raw-based algorithm (Hees). In addition, we performed a sensitivity analysis to provide insight into the relationship between the algorithms’ hyperparameters and classification performance, as well as to generate tuned hyperparameter values to better detect episodes of wear and non-wear time. We used machine learning to construct a gold-standard dataset by combining two accelerometers and electrocardiogram recordings. The Hecht and Troiano algorithms achieved poor classification performance, while Choi exhibited moderate performance. Meanwhile, Hees outperformed all epoch-based algorithms. The sensitivity analysis and hyperparameter tuning revealed that all algorithms were able to achieve increased classification performance by employing larger intervals and windows, while more stringently defining artificial movement. These classification gains were associated with the ability to lower the false positives (type I error) and do not necessarily indicate a more accurate detection of the total non-wear time. Moreover, our results indicate that with tuned hyperparameters, epoch-based non-wear algorithms are able to perform just as well as raw-based non-wear algorithms with respect to their ability to correctly detect true wear and non-wear episodes.

FilterK: A new outlier detection method for k-means clustering of physical activity

In this paper, a new algorithm denoted as FilterK is proposed for improving the purity of k-means derived physical activity clusters by reducing outlier influence. We applied it to physical activity data obtained with body-worn accelerometers and clustered using k-means. We compared its performance with three existing outlier detection methods: Local Outlier Factor, Isolation Forests and KNN using the ground truth (class labels), average cluster and event purity (ACEP). FilterK provided comparable gains in ACEP (0.581 → 0.596 compared to 0.580-0.617) whilst removing a lower number of outliers than the other methods (4% total dataset size vs 10% to achieve this ACEP). The main focus of our new outlier detection method is to improve the cluster purities of physical activity accelerometer data, but we also suggest it may be potentially applied to other types of dataset captured by k-means clustering. We demonstrate our method using a k-means model trained on two independent accelerometer datasets (training n = 90) and re-applied to an independent dataset (test n = 41). Labelled physical activities include lying down, sitting, standing, household chores, walking (laboratory and non-laboratory based), stairs and running. This type of clustering algorithm could be used to assist with identifying optimal physical activity patterns for health.

Prediction of Diabetic Foot Ulceration: The Value of Using Microclimate Sensor Arrays

BACKGROUND

Accurately predicting the risk of diabetic foot ulceration (DFU) could dramatically reduce the enormous burden of chronic wound management and amputation. Yet, the current prognostic models are unable to precisely predict DFU events. Typically, efforts have focused on individual factors like temperature, pressure, or shear rather than the overall foot microclimate.

METHODS

A systematic review was conducted by searching PubMed reports with no restrictions on start date covering the literature published until February 20, 2019 using relevant keywords, including temperature, pressure, shear, and relative humidity. We review the use of these variables as predictors of DFU, highlighting gaps in our current understanding and suggesting which specific features should be combined to develop a real-time microclimate prognostic model.

RESULTS

The current prognostic models rely either solely on contralateral temperature, pressure, or shear measurement; these parameters, however, rarely reach 50% specificity in relation to DFU. There is also considerable variation in methodological investigation, anatomical sensor configuration, and resting time prior to temperature measurements (5-20 minutes). Few studies have considered relative humidity and mean skin resistance.

CONCLUSION

Very limited evidence supports the use of single clinical parameters in predicting the risk of DFU. We suggest that the microclimate as a whole should be considered to predict DFU more effectively and suggest nine specific features which appear to be implicated for further investigation. Technology supports real-time in-shoe data collection and wireless transmission, providing a potentially rich source of data to better predict the risk of DFU.

Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data

Few methods for classifying physical activity from accelerometer data have been tested using an independent dataset for cross-validation, and even fewer using multiple independent datasets. The aim of this study was to evaluate whether unsupervised machine learning was a viable approach for the development of a reusable clustering model that was generalisable to independent datasets. We used two labelled adult laboratory datasets to generate a k-means clustering model. To assess its generalised application, we applied the stored clustering model to three independent labelled datasets: two laboratory and one free-living. Based on the development labelled data, the ten clusters were collapsed into four activity categories: sedentary, standing/mixed/slow ambulatory, brisk ambulatory, and running. The percentages of each activity type contained in these categories were 89%, 83%, 78%, and 96%, respectively. In the laboratory independent datasets, the consistency of activity types within the clusters dropped, but remained above 70% for the sedentary clusters, and 85% for the running and ambulatory clusters. Acceleration features were similar within each cluster across samples. The clusters created reflected activity types known to be associated with health and were reasonably robust when applied to diverse independent datasets. This suggests that an unsupervised approach is potentially useful for analysing free-living accelerometer data.

Calibration and Validation of the Youth Activity Profile as a Physical Activity and Sedentary Behaviour Surveillance Tool for English Youth

Self-reported youth physical activity (PA) is typically overestimated. We aimed to calibrate and validate a self-report tool among English youth. Four-hundred-and-two participants (aged 9-16 years; 212 boys) wore SenseWear Armband Mini devices (SWA) for eight days and completed the self-report Youth Activity Profile (YAP) on the eighth day. Calibration algorithms for temporally matched segments were generated from the YAP data using quantile regression. The algorithms were applied in an independent cross-validation sample, and student- and school-level agreement were assessed. The utility of the YAP algorithms to assess compliance to PA guidelines was also examined. The school-level bias for the YAP estimates of in-school, out-of-school, and weekend moderate-to-vigorous PA (MVPA) were 17.2 (34.4), 31.6 (14.0), and -4.9 (3.6) min·week^-1, respectively. Out-of-school sedentary behaviour (SB) was over-predicted by 109.2 (11.8) min·week^-1. Predicted YAP values were within 15%-20% equivalence of the SWA estimates. The classification accuracy of the YAP MVPA estimates for compliance to 60 min·day^-1 and 30 min·school-day^-1 MVPA recommendations were 91%/37% and 89%/57% sensitivity/specificity, respectively. The YAP generated robust school-level estimates of MVPA and SB and has potential for surveillance to monitor compliance with PA guidelines. The accuracy of the YAP may be further improved through research with more representative UK samples to enhance the calibration process and to refine the resultant algorithms.