Using Wearable Activity Type Detection to Improve Physical Activity Energy Expenditure Estimation

Nutrition and Golf Performance: A Systematic Scoping Review

BACKGROUND

Golf is played both recreationally and professionally by approximately 66.6 million people worldwide. Despite the potential for nutrition to influence golf performance, research in this area is somewhat limited.

OBJECTIVE

To identify the existing literature regarding nutrition and golf and where the current research gaps lie.

DESIGN

Scoping review. Online databases were used to retrieve data from 2003 to the present day.

DATA SOURCES

A three-step search strategy identified relevant primary and secondary articles as well as grey literature. Published and unpublished articles in the English language, identified by searching electronic databases (ProQuest Central, Web of Science, Scopus, SPORTDiscus and PubMed) and reference searching.

REVIEW METHODS

Relevant identified studies were screened for final inclusion. Data were extracted using a standardised tool to create a descriptive analysis and a thematic summary. In summary, studies were included if they focused on nutrition, hydration, energy requirements, supplements, or body composition in relation to golf.

RESULTS AND DISCUSSION

Our initial search found 3616 relevant articles. Eighty-two of these articles were included for the scoping review. Nutrition has the potential to impact golf performance in areas including the maintenance of energy levels, cognitive function, and body composition. Currently, there is limited research available discussing the effects of nutrition interventions related specifically to golf performance.

CONCLUSION

This scoping review highlights that more work is needed to provide golfers and practitioners with golf-specific nutrition research. The key areas for future golf-specific nutrition research include nutrition on cognitive performance, body composition, energy requirements, supplementation, and the potential role of nutrition for the travelling golfer. Systematic reviews could also be used to identify future priorities for nutrition and golf research.

A multidevice and multimodal dataset for human energy expenditure estimation using wearable devices

We present a multi-device and multi-modal dataset, called WEEE, collected from 17 participants while they were performing different physical activities. WEEE contains: (1) sensor data collected using seven wearable devices placed on four body locations (head, ear, chest, and wrist); (2) respiratory data collected with an indirect calorimeter serving as ground-truth information; (3) demographics and body composition data (e.g., fat percentage); (4) intensity level and type of physical activities, along with their corresponding metabolic equivalent of task (MET) values; and (5) answers to questionnaires about participants' physical activity level, diet, stress and sleep. Thanks to the diversity of sensors and body locations, we believe that the dataset will enable the development of novel human energy expenditure (EE) estimation techniques for a diverse set of application scenarios. EE refers to the amount of energy an individual uses to maintain body functions and as a result of physical activity. A reliable estimate of people's EE thus enables computing systems to make inferences about users' physical activity and help them promoting a healthier lifestyle.

Sensors Capabilities, Performance, and Use of Consumer Sleep Technology

Sleep is crucial for the proper functioning of bodily systems and for cognitive and emotional processing. Evidence indicates that sleep is vital for health, well-being, mood, and performance. Consumer sleep technologies (CSTs), such as multisensory wearable devices, have brought attention to sleep and there is growing interest in using CSTs in research and clinical applications. This article reviews how CSTs can process information about sleep, physiology, and environment. The growing number of sensors in wearable devices and the meaning of the data collected are reviewed. CSTs have the potential to provide opportunities to measure sleep and sleep-related physiology on a large scale.

Estimating Upper-Limb Impairment Level in Stroke Survivors Using Wearable Inertial Sensors and a Minimally-Burdensome Motor Task

Upper-limb paresis is the most common motor impairment post stroke. Current solutions to automate the assessment of upper-limb impairment impose a number of critical burdens on patients and their caregivers that preclude frequent assessment. In this work, we propose an approach to estimate upper-limb impairment in stroke survivors using two wearable inertial sensors, on the wrist and the sternum, and a minimally-burdensome motor task. Twenty-three stroke survivors with no, mild, or moderate upper-limb impairment performed two repetitions of one-to-two minute-long continuous, random (i.e., patternless), voluntary upper-limb movements spanning the entire range of motion. The three-dimensional time-series of upper-limb movements were segmented into a series of one-dimensional submovements by employing a unique movement decomposition technique. An unsupervised clustering algorithm and a supervised regression model were used to estimate Fugl-Meyer Assessment (FMA) scores based on features extracted from these submovements. Our regression model estimated FMA scores with a normalized root mean square error of 18.2% ( r²=0.70 ) and needed as little as one minute of movement data to yield reasonable estimation performance. These results support the possibility of frequently monitoring stroke survivors' rehabilitation outcomes, ultimately enabling the development of individually-tailored rehabilitation programs.

Physical Workload Tracking Using Human Activity Recognition with Wearable Devices

In this work, authors address workload computation combining human activity recognition and heart rate measurements to establish a scalable framework for health at work and fitness-related applications. The proposed architecture consists of two wearable sensors: one for motion, and another for heart rate. The system employs machine learning algorithms to determine the activity performed by a user, and takes a concept from ergonomics, the Frimat's score, to compute the corresponding physical workload from measured heart rate values providing in addition a qualitative description of the workload. A random forest activity classifier is trained and validated with data from nine subjects, achieving an accuracy of 97.5%. Then, tests with 20 subjects show the reliability of the activity classifier, which keeps an accuracy up to 92% during real-time testing. Additionally, a single-subject twenty-day physical workload tracking case study evinces the system capabilities to detect body adaptation to a custom exercise routine. The proposed system enables remote and multi-user workload monitoring, which facilitates the job for experts in ergonomics and workplace health.

Wrist accelerometer shape feature derivation methods for assessing activities of daily living

Background

There has been an increasing interest in understanding the usefulness of wrist-based accelerometer data for physical activity (PA) assessment due to the ease of use and higher user compliance than other body placements. PA assessment studies have relied on machine learning methods which take accelerometer data in forms of variables, or feature vectors.

Methods

In this work, we introduce automated shape feature derivation methods to transform epochs of accelerometer data into feature vectors. As the first step, recurring patterns in the collected data are identified and placed in a codebook. Similarities between epochs of accelerometer data and codebook’s patterns are the basis of feature calculations. In this paper, we demonstrate supervised and unsupervised approaches to learn codebooks. We evaluated these methods and compared them with the standard statistical measures for PA assessment. The experiments were performed on 146 participants who wore an ActiGraph GT3X+ accelerometer on the right wrist and performed 33 activities of daily living.

Results

Our evaluations show that the shape feature derivation methods were able to perform comparably with the standard wrist model (F1-score: 0.89) for identifying sedentary PAs (F1-scores of 0.86 and 0.85 for supervised and unsupervised methods, respectively). This was also observed for identifying locomotion activities (F1-scores: 0.87, 0.83, and 0.81 for the standard wrist, supervised, unsupervised models, respectively). All the wrist models were able to estimate energy expenditure required for PAs with low error (rMSE: 0.90, 0.93, and 0.90 for the standard wrist, supervised, and unsupervised models, respectively).

Conclusion

The automated shape feature derivation methods offer insights into the performed activities by providing a summary of repeating patterns in the accelerometer data. Furthermore, they could be used as efficient alternatives (or additions) for manually engineered features, especially important for cases where the latter fail to provide sufficient information to machine learning methods for PA assessment.

Body Sensor Networks Powered by an NFC-Coupled Smartphone in the Pocket

This paper proposes a body sensor network (BSN) on clothing that is wirelessly powered by a smartphone in a pocket. The network consists of a host device and multiple sensor nodes, which are distributed on a wear and are electrically connected with conductive threads. The smartphone with a built-in near field communication (NFC) feature powers the host, which is fixed at the pocket. These devices are wired to a special cloth embroidered with conductive threads by using a special connector consisting of a pin & socket without one-to-one wiring. In the proposed BSN, the host device and the smartphone are coupled via NFC radio within the pocket. Energy harvesting with NFC radio wave requires maintaining antennas within several centimeters to obtain enough power. Positioning and fixing of the smartphone is required within the pocket. A proposed host device can expand the range of energy harvesting by using multiple antennas and a power aggregation circuit. The experimental results demonstrate the feasibility of the batteryless BSNs system.

Cross-Modal Health State Estimation

Individuals create and consume more diverse data about themselves today than any time in history. Sources of this data include wearable devices, images, social media, geo-spatial information and more. A tremendous opportunity rests within cross-modal data analysis that leverages existing domain knowledge methods to understand and guide human health. Especially in chronic diseases, current medical practice uses a combination of sparse hospital based biological metrics (blood tests, expensive imaging, etc.) to understand the evolving health status of an individual. Future health systems must integrate data created at the individual level to better understand health status perpetually, especially in a cybernetic framework. In this work we fuse multiple user created and open source data streams along with established biomedical domain knowledge to give two types of quantitative state estimates of cardiovascular health. First, we use wearable devices to calculate cardiorespiratory fitness (CRF), a known quantitative leading predictor of heart disease which is not routinely collected in clinical settings. Second, we estimate inherent genetic traits, living environmental risks, circadian rhythm, and biological metrics from a diverse dataset. Our experimental results on 24 subjects demonstrate how multi-modal data can provide personalized health insight. Understanding the dynamic nature of health status will pave the way for better health based recommendation engines, better clinical decision making and positive lifestyle changes.

Advances and Controversies in Diet and Physical Activity Measurement in Youth

Technological advancements in the past decades have improved dietary intake and physical activity measurements. This report reviews current developments in dietary intake and physical activity assessment in youth. Dietary intake assessment has relied predominantly on self-report or image-based methods to measure key aspects of dietary intake (e.g., food types, portion size, eating occasion), which are prone to notable methodologic (e.g., recall bias) and logistic (e.g., participant and researcher burden) challenges. Although there have been improvements in automatic eating detection, artificial intelligence, and sensor-based technologies, participant input is often needed to verify food categories and portions. Current physical activity assessment methods, including self-report, direct observation, and wearable devices, provide researchers with reliable estimations for energy expenditure and bodily movement. Recent developments in algorithms that incorporate signals from multiple sensors and technology-augmented self-reporting methods have shown preliminary efficacy in measuring specific types of activity patterns and relevant contextual information. However, challenges in detecting resistance (e.g., in resistance training, weight lifting), prolonged physical activity monitoring, and algorithm (non)equivalence remain to be addressed. In summary, although dietary intake assessment methods have yet to achieve the same validity and reliability as physical activity measurement, recent developments in wearable technologies in both arenas have the potential to improve current assessment methods.

THEME INFORMATION

This article is part of a theme issue entitled Innovative Tools for Assessing Diet and Physical Activity for Health Promotion, which is sponsored by the North American branch of the International Life Sciences Institute.

Do moderate- to vigorous-intensity accelerometer count thresholds correspond to relative moderate- to vigorous-intensity physical activity?

We aimed to predict % maximal oxygen consumption at absolute accelerometer thresholds and to estimate and compare durations of objective physical activity (PA) among body mass index (BMI) categories using thresholds that account for cardiorespiratory fitness. Eight hundred twenty-eight adults (53.5% male; age, 33.9 ± 0.3 years) from the National Health and Nutrition Examination Survey 2003-2004 were analyzed. Metabolic equivalent values at absolute thresholds were converted to percentage of maximal oxygen consumption, and accelerometer counts corresponding to 40% or 60% maximal oxygen consumption were determined using 4 energy expenditure prediction equations. Absolute thresholds underestimated PA intensity for all adults; however, because of lower fitness, individuals with overweight and obesity work at significantly higher percentage of maximal oxygen consumption at the absolute thresholds and require significantly lower accelerometer counts to reach relative moderate and vigorous PA intensities compared with those with normal weight (P < 0.05). However, moderate-to-vigorous physical activity (MVPA) durations were shorter when using relative thresholds compared with absolute thresholds (in all BMI groups, P < 0.05), and they were shorter among individuals with obesity compared with those with normal weight when using relative thresholds (P < 0.05). Regardless of the thresholds used, a greater proportion of individuals with normal weight met the PA guideline of 150 min·week^-1 of MVPA compared with individuals with obesity (absolute: 21.3% vs 6.7%; Yngve: 4.0% vs 0.2%; Swartz: 10.7% vs 3.9%; Hendelman: 4.7% vs 0.2%; Freedson: 6.4% vs 0.5%; P < 0.05). Current absolute thresholds of accelerometry-derived PA may overestimate MVPA for all BMI categories when compared with relative thresholds that account for cardiorespiratory fitness. Given the large variability in our results, more work is needed to better understand how to use accelerometers for evaluating PA at the population level.

Fusion of Heart Rate, Respiration and Motion Measurements from a Wearable Sensor System to Enhance Energy Expenditure Estimation

This paper presents a new method that integrates heart rate, respiration, and motion information obtained from a wearable sensor system to estimate energy expenditure. The system measures electrocardiography, impedance pneumography, and acceleration from upper and lower limbs. A multilayer perceptron neural network model was developed, evaluated, and compared to two existing methods, with data from 11 subjects (mean age, 27 years, range, 21⁻65 years) who performed a 3-h protocol including submaximal tests, simulated work tasks, and periods of rest. Oxygen uptake was measured with an indirect calorimeter as a reference, with a time resolution of 15 s. When compared to the reference, the new model showed a lower mean absolute error (MAE = 1.65 mL/kg/min, R² = 0.92) than the two existing methods, i.e., the flex-HR method (MAE = 2.83 mL/kg/min, R² = 0.75), which uses only heart rate, and arm-leg HR+M method (MAE = 2.12 mL/kg/min, R² = 0.86), which uses heart rate and motion information. As indicated, this new model may, in combination with a wearable system, be useful in occupational and general health applications.

Classifier Personalization for Activity Recognition Using Wrist Accelerometers

Intersubject variability in accelerometer-based activity recognition may significantly affect classification accuracy, limiting a reliable extension of methods to new users. In this paper, we propose an approach for personalizing classification rules to a single person. We demonstrate that the method improves activity detection from wrist-worn accelerometer data on a four-class recognition problem of interest to the exercise science community, where classes are ambulation, cycling, sedentary, and other. We extend a previously published activity classification method based on support vector machines so that it estimates classification uncertainty. Uncertainty is used to drive data label requests from the user, and the resulting label information is used to update the classifier. Two different datasets-one from 33 adults with 26 activity types, and another from 20 youth with 23 activity types-were used to evaluate the method using leave-one-subject-out and leave-one-group-out cross validation. The new method improved overall recognition accuracy up to 11% on average, with some large person-specific improvements (ranging from -2% to +36%). The proposed method is suitable for online implementation supporting real-time recognition systems.

Data correction for seven activity trackers based on regression models

Using an activity tracker for measuring activity-related parameters, e.g. steps and energy expenditure (EE), can be very helpful in assisting a person's fitness improvement. Unlike the measuring of number of steps, an accurate EE estimation requires additional personal information as well as accurate velocity of movement, which is hard to achieve due to inaccuracy of sensors. In this paper, we have evaluated regression-based models to improve the precision for both steps and EE estimation. For this purpose, data of seven activity trackers and two reference devices was collected from 20 young adult volunteers wearing all devices at once in three different tests, namely 60-minute office work, 6-hour overall activity and 60-minute walking. Reference data is used to create regression models for each device and relative percentage errors of adjusted values are then statistically compared to that of original values. The effectiveness of regression models are determined based on the result of a statistical test. During a walking period, EE measurement was improved in all devices. The step measurement was also improved in five of them. The results show that improvement of EE estimation is possible only with low-cost implementation of fitting model over the collected data e.g. in the app or in corresponding service back-end.

Automatic Identification of Physical Activity Intensity and Modality from the Fusion of Accelerometry and Heart Rate Data

Background: Physical activity (PA) is essential to prevent and to treat a variety of chronic diseases. The automated detection and quantification of PA over time empowers lifestyle interventions, facilitating reliable exercise tracking and data-driven counseling.

Methods: We propose and compare various combinations of machine learning (ML) schemes for the automatic classification of PA from multi-modal data, simultaneously captured by a biaxial accelerometer and a heart rate (HR) monitor. Intensity levels (low / moderate / vigorous) were recognized, as well as for vigorous exercise, its modality (sustained aerobic / resistance / mixed). In to -tal, 178.63 h of data about PA intensity (65.55 % low / 18.96 % moderate / 15.49 % vigorous) and 17.00 h about modality were collected in two experiments: one in free- living conditions, another in a fitness center under controlled protocols. The structure used for automatic classification comprised: a) definition of 42 time-domain signal features, b) dimensionality reduction, c) data clustering, and d) temporal filtering to exploit time redundancy by means of a Hidden Markov Model (HMM). Four dimensionality reduction techniques and four clustering algorithms were studied. In order to cope with class imbalance in the dataset, a custom performance metric was defined to aggregate recognition accuracy, precision and recall.

Results: The best scheme, which comprised a projection through Linear Discriminant Ana -lysis (LDA) and k-means clustering, was evaluated in leave-one-subject-out cross-validation; notably outperforming the standard industry procedures for PA intensity classification: score 84.65 %, versus up to 63.60 %. Errors tended to be brief and to appear around transients.

Conclusions: The application of ML techniques for pattern identification and temporal filtering allowed to merge accelerometry and HR data in a solid manner, and achieved markedly better recognition performances than the standard methods for PA intensity estimation.

A method to estimate free-living active and sedentary behavior from an accelerometer

INTRODUCTION

Methods to estimate physical activity (PA) and sedentary behavior (SB) from wearable monitors need to be validated in free-living settings.

PURPOSE

The purpose of this study was to develop and validate two novel machine-learning methods (Sojourn-1 Axis [soj-1x] and Sojourn-3 Axis [soj-3x]) in a free-living setting.

METHODS

Participants were directly observed in their natural environment for 10 consecutive hours on three separate occasions. Physical activity and SB estimated from soj-1x, soj-3x, and a neural network previously calibrated in the laboratory (lab-nnet) were compared with direct observation.

RESULTS

Compared with lab-nnet, soj-1x and soj-3x improved estimates of MET-hours (lab-nnet: % bias [95% confidence interval] = 33.1 [25.9 to 40.4], root-mean-square error [RMSE] = 5.4 [4.6-6.2]; soj-1x: % bias = 1.9 [-2.0 to 5.9], RMSE = 1.0 [0.6 to 1.3]; soj-3x: % bias = 3.4 [0.0 to 6.7], RMSE = 1.0 [0.6 to 1.5]) and minutes in different intensity categories {lab-nnet: % bias = -8.2 (sedentary), -8.2 (light), and 72.8 (moderate-to-vigorous PA [MVPA]); soj-1x: % bias = 8.8 (sedentary), -18.5 (light), and -1.0 (MVPA); soj-3x: % bias = 0.5 (sedentary), -0.8 (light), and -1.0 (MVPA)}. Soj-1x and soj-3x also produced accurate estimates of guideline minutes and breaks from sedentary time.

CONCLUSIONS

Compared with the lab-nnet algorithm, soj-1x and soj-3x improved the accuracy and precision in estimating free-living MET-hours, sedentary time, and time spent in light-intensity activity and MVPA. In addition, soj-3x is superior to soj-1x in differentiating SB from light-intensity activity.

Exercise Performance Measurement with Smartphone Embedded Sensor for Well-Being Management

Regular physical activity reduces the risk of many diseases and improves physical and mental health. However, physical inactivity is widespread globally. Improving physical activity levels is a global concern in well-being management. Exercise performance measurement systems have the potential to improve physical activity by providing feedback and motivation to users. We propose an exercise performance measurement system for well-being management that is based on the accumulated activity effective index (AAEI) and incorporates a smartphone-embedded sensor. The proposed system generates a numeric index that is based on users' exercise performance: their level of physical activity and number of days spent exercising. The AAEI presents a clear number that can serve as a useful feedback and goal-setting tool. We implemented the exercise performance measurement system by using a smartphone and conducted experiments to assess the feasibility of the system and investigated the user experience. We recruited 17 participants for validating the feasibility of the measurement system and a total of 35 participants for investigating the user experience. The exercise performance measurement system showed an overall precision of 88% in activity level estimation. Users provided positive feedback about their experience with the exercise performance measurement system. The proposed system is feasible and has a positive effective on well-being management.

Estimation of Energy Expenditure Using a Patch-Type Sensor Module with an Incremental Radial Basis Function Neural Network

Conventionally, indirect calorimetry has been used to estimate oxygen consumption in an effort to accurately measure human body energy expenditure. However, calorimetry requires the subject to wear a mask that is neither convenient nor comfortable. The purpose of our study is to develop a patch-type sensor module with an embedded incremental radial basis function neural network (RBFNN) for estimating the energy expenditure. The sensor module contains one ECG electrode and a three-axis accelerometer, and can perform real-time heart rate (HR) and movement index (MI) monitoring. The embedded incremental network includes linear regression (LR) and RBFNN based on context-based fuzzy c-means (CFCM) clustering. This incremental network is constructed by building a collection of information granules through CFCM clustering that is guided by the distribution of error of the linear part of the LR model.

Comparative evaluation of features and techniques for identifying activity type and estimating energy cost from accelerometer data

Wearable accelerometers can be used to objectively assess physical activity. However, the accuracy of this assessment depends on the underlying method used to process the time series data obtained from accelerometers. Several methods have been proposed that use this data to identify the type of physical activity and estimate its energy cost. Most of the newer methods employ some machine learning technique along with suitable features to represent the time series data. This paper experimentally compares several of these techniques and features on a large dataset of 146 subjects doing eight different physical activities wearing an accelerometer on the hip. Besides features based on statistics, distance based features and simple discrete features straight from the time series were also evaluated. On the physical activity type identification task, the results show that using more features significantly improve results. Choice of machine learning technique was also found to be important. However, on the energy cost estimation task, choice of features and machine learning technique were found to be less influential. On that task, separate energy cost estimation models trained specifically for each type of physical activity were found to be more accurate than a single model trained for all types of physical activities.

Accelerometry-based Recognition of the Placement Sites of a Wearable Sensor

This work describes an automatic method to recognize the position of an accelerometer worn on five different parts of the body: ankle, thigh, hip, arm and wrist from raw accelerometer data. Automatic detection of body position of a wearable sensor would enable systems that allow users to wear sensors flexibly on different body parts or permit systems that need to automatically verify sensor placement. The two-stage location detection algorithm works by first detecting time periods during which candidates are walking (regardless of where the sensor is positioned). Then, assuming that the data refer to walking, the algorithm detects the position of the sensor. Algorithms were validated on a dataset that is substantially larger than in prior work, using a leave-one-subject-out cross-validation approach. Correct walking and placement recognition were obtained for 97.4% and 91.2% of classified data windows, respectively.

A Comparative Study on the Suitability of Smartphones and IMU for Mobile, Unsupervised Energy Expenditure Calculi

The metabolic equivalent of task (MET) is currently the most used indicator for measuring the energy expenditure (EE) of a physical activity (PA) and has become an important measure for determining and supervising a person’s state of health. The use of new devices which are capable of measuring inertial movements by means of built-in accelerometers enable the PA to be measured objectively on the basis of the reckoning of “counts”. These devices are also known as inertial measurement units (IMUs) and each count is an aggregated value indicating the intensity of a movement and can be used in conjunction with other parameters to determine the MET rate of a particular physical activity and thus it’s associated EE. Various types of inertial devices currently exist that enable count calculus and physical activity to be monitored. The advent of mobile devices, such as smartphones, with empowered computation capabilities and integrated inertial sensors, has enabled EE to be measure in a distributed, ubiquitous and natural way, thereby overcoming the reluctance of users and practitioners associated with in-lab studies. From the point of view of the process analysis and infrastructure needed to manage data from inertial devices, there are also various differences in count computing: extra devices are required, out-of-device processing, etc. This paper presents a study to discover whether the estimation of energy expenditure is dependent on the accelerometer of the device used in measurements and to discover the suitability of each device for performing certain physical activities. In order to achieve this objective, we have conducted several experiments with different subjects on the basis of the performance of various daily activities with different smartphones and IMUs.

Early Detection of Physical Activity for People With Type 1 Diabetes Mellitus

BACKGROUND

Early detection of exercise in individuals with type 1 diabetes mellitus (T1DM) may allow changes in therapy to prevent hypoglycemia. Currently there is limited experience with automated methods that detect the onset and end of exercise in this population. We sought to develop a novel method to quickly and reliably detect the onset and end of exercise in these individuals before significant changes in blood glucose (BG) occur.

METHODS

Sixteen adults with T1DM were studied as outpatients using a diary, accelerometer, heart rate monitor, and continuous glucose monitor for 2 days. These data were used to develop a principal component analysis based exercise detection method. Subjects also performed 60 and 30 minute exercise sessions at 30% and 50% predicted heart rate reserve (HRR), respectively. The detection method was applied to the exercise sessions to determine how quickly the detection of start and end of exercise occurred relative to change in BG.

RESULTS

Mild 30% HRR and moderate 50% HRR exercise onset was identified in 6 ± 3 and 5 ± 2 (mean ± SD) minutes, while completion was detected in 3 ± 8 and 6 ± 5 minutes, respectively. BG change from start of exercise to detection time was 1 ± 6 and -1 ± 3 mg/dL, and, from the end of exercise to detection time was 6 ± 4 and -17 ± 13 mg/dL, respectively, for the 2 exercise sessions. False positive and negative ratios were 4 ± 2% and 21 ± 22%.

CONCLUSIONS

The novel method for exercise detection identified the onset and end of exercise in approximately 5 minutes, with an average BG change of only -6 mg/dL.

Methods to estimate aspects of physical activity and sedentary behavior from high-frequency wrist accelerometer measurements

This investigation developed models to estimate aspects of physical activity and sedentary behavior from three-axis high-frequency wrist-worn accelerometer data. The models were developed and tested on 20 participants (n = 10 males, n = 10 females, mean age = 24.1, mean body mass index = 23.9), who wore an ActiGraph GT3X+ accelerometer on their dominant wrist and an ActiGraph GT3X on the hip while performing a variety of scripted activities. Energy expenditure was concurrently measured by a portable indirect calorimetry system. Those calibration data were then used to develop and assess both machine-learning and simpler models with fewer unknown parameters (linear regression and decision trees) to estimate metabolic equivalent scores (METs) and to classify activity intensity, sedentary time, and locomotion time. The wrist models, applied to 15-s windows, estimated METs [random forest: root mean squared error (rSME) = 1.21 METs, hip: rMSE = 1.67 METs] and activity intensity (random forest: 75% correct, hip: 60% correct) better than a previously developed model that used counts per minute measured at the hip. In a separate set of comparisons, the simpler decision trees classified activity intensity (random forest: 75% correct, tree: 74% correct), sedentary time (random forest: 96% correct, decision tree: 97% correct), and locomotion time (random forest: 99% correct, decision tree: 96% correct) nearly as well or better than the machine-learning approaches. Preliminary investigation of the models' performance on two free-living people suggests that they may work well outside of controlled conditions.

Patient-centered activity monitoring in the self-management of chronic health conditions

BACKGROUND

As activity tracking devices become smaller, cheaper, and more consumer-accessible, they will be used more extensively across a wide variety of contexts. The expansion of activity tracking and personal data collection offers the potential for patient engagement in the management of chronic diseases. Consumer wearable devices for activity tracking have shown promise in post-surgery recovery in cardiac patients, pulmonary rehabilitation, and activity counseling in diabetic patients, among others. Unfortunately, the data generated by wearable devices is seldom integrated into programmatic self-management chronic disease regimens. In addition, there is lack of evidence supporting sustained use or effects on health outcomes, as studies have primarily focused on establishing the feasibility of monitoring activity and the association of measured activity with short-term benefits.

DISCUSSION

Monitoring devices can make a direct and real-time impact on self-management, but the validity and reliability of measurements need to be established. In order for patients to become engaged in wearable data gathering, key patient-centered issues relating to usefulness in care, motivation, the safety and privacy of information, and clinical integration need to be addressed. Because the successful usage of wearables requires an ability to comprehend and utilize personal health data, the user experience should account for individual differences in numeracy skills and apply evidence-based behavioral science principles to promote continued engagement.

SUMMARY

Activity monitoring has the potential to engage patients as advocates in their personalized care, as well as offer health care providers real world assessments of their patients' daily activity patterns. This potential will be realized as the voice of the chronic disease patients is accounted for in the design of devices, measurements are validated against existing clinical assessments, devices become part of the treatment 'prescription', behavior change programs are used to engage patients in self-management, and best practices for clinical integration are defined.

Hierarchical approaches to estimate energy expenditure using phone-based accelerometers

Physical inactivity is linked with increase in risk of cancer, heart disease, stroke, and diabetes. Walking is an easily available activity to reduce sedentary time. Objective methods to accurately assess energy expenditure from walking that is normalized to an individual would allow tailored interventions. Current techniques rely on normalization by weight scaling or fitting a polynomial function of weight and speed. Using the example of steady-state treadmill walking, we present a set of algorithms that extend previous work to include an arbitrary number of anthropometric descriptors. We specifically focus on predicting energy expenditure using movement measured by mobile phone-based accelerometers. The models tested include nearest neighbor models, weight-scaled models, a set of hierarchical linear models, multivariate models, and speed-based approaches. These are compared for prediction accuracy as measured by normalized average root mean-squared error across all participants. Nearest neighbor models showed highest errors. Feature combinations corresponding to sedentary energy expenditure, sedentary heart rate, and sex alone resulted in errors that were higher than speed-based models and nearest-neighbor models. Size-based features such as BMI, weight, and height produced lower errors. Hierarchical models performed better than multivariate models when size-based features were used. We used the hierarchical linear model to determine the best individual feature to describe a person. Weight was the best individual descriptor followed by height. We also test models for their ability to predict energy expenditure with limited training data. Hierarchical models outperformed personal models when a low amount of training data were available. Speed-based models showed poor interpolation capability, whereas hierarchical models showed uniform interpolation capabilities across speeds.

A random forest classifier for the prediction of energy expenditure and type of physical activity from wrist and hip accelerometers

Wrist accelerometers are being used in population level surveillance of physical activity (PA) but more research is needed to evaluate their validity for correctly classifying types of PA behavior and predicting energy expenditure (EE). In this study we compare accelerometers worn on the wrist and hip, and the added value of heart rate (HR) data, for predicting PA type and EE using machine learning. Forty adults performed locomotion and household activities in a lab setting while wearing three ActiGraph GT3X+ accelerometers (left hip, right hip, non-dominant wrist) and a HR monitor (Polar RS400). Participants also wore a portable indirect calorimeter (COSMED K4b2), from which EE and metabolic equivalents (METs) were computed for each minute. We developed two predictive models: a random forest classifier to predict activity type and a random forest of regression trees to estimate METs. Predictions were evaluated using leave-one-user-out cross-validation. The hip accelerometer obtained an average accuracy of 92.3% in predicting four activity types (household, stairs, walking, running), while the wrist accelerometer obtained an average accuracy of 87.5%. Across all 8 activities combined (laundry, window washing, dusting, dishes, sweeping, stairs, walking, running), the hip and wrist accelerometers obtained average accuracies of 70.2% and 80.2% respectively. Predicting METs using the hip or wrist devices alone obtained root mean square errors (rMSE) of 1.09 and 1.00 METs per 6 min bout, respectively. Including HR data improved MET estimation, but did not significantly improve activity type classification. These results demonstrate the validity of random forest classification and regression forests for PA type and MET prediction using accelerometers. The wrist accelerometer proved more useful in predicting activities with significant arm movement, while the hip accelerometer was superior for predicting locomotion and estimating EE.

Estimating energy expenditure using body-worn accelerometers: a comparison of methods, sensors number and positioning

Several methods to estimate energy expenditure (EE) using body-worn sensors exist; however, quantifications of the differences in estimation error are missing. In this paper, we compare three prevalent EE estimation methods and five body locations to provide a basis for selecting among methods, sensors number, and positioning. We considered 1) counts-based estimation methods, 2) activity-specific estimation methods using METs lookup, and 3) activity-specific estimation methods using accelerometer features. The latter two estimation methods utilize subsequent activity classification and EE estimation steps. Furthermore, we analyzed accelerometer sensors number and on-body positioning to derive optimal EE estimation results during various daily activities. To evaluate our approach, we implemented a study with 15 participants that wore five accelerometer sensors while performing a wide range of sedentary, household, lifestyle, and gym activities at different intensities. Indirect calorimetry was used in parallel to obtain EE reference data. Results show that activity-specific estimation methods using accelerometer features can outperform counts-based methods by 88% and activity-specific methods using METs lookup for active clusters by 23%. No differences were found between activity-specific methods using METs lookup and using accelerometer features for sedentary clusters. For activity-specific estimation methods using accelerometer features, differences in EE estimation error between the best combinations of each number of sensors (1 to 5), analyzed with repeated measures ANOVA, were not significant. Thus, we conclude that choosing the best performing single sensor does not reduce EE estimation accuracy compared to a five sensors system and can reliably be used. However, EE estimation errors can increase up to 80% if a nonoptimal sensor location is chosen.

Hierarchical Linear Models for Energy Prediction using Inertial Sensors: A Comparative Study for Treadmill Walking

Walking is a commonly available activity to maintain a healthy lifestyle. Accurately tracking and measuring calories expended during walking can improve user feedback and intervention measures. Inertial sensors are a promising measurement tool to achieve this purpose. An important aspect in mapping inertial sensor data to energy expenditure is the question of normalizing across physiological parameters. Common approaches such as weight scaling require validation for each new population. An alternative is to use a hierarchical approach to model subject-specific parameters at one level and cross-subject parameters connected by physiological variables at a higher level. In this paper, we evaluate an inertial sensor-based hierarchical model to measure energy expenditure across a target population. We first determine the optimal movement and physiological features set to represent data. Periodicity based features are more accurate (p<0.1 per subject) when generalizing across populations. Weight is the most accurate parameter (p<0.1 per subject) measured as percentage prediction error. We also compare the hierarchical model with a subject-specific regression model and weight exponent scaled models. Subject-specific models perform significantly better (p<0.1 per subject) than weight exponent scaled models at all exponent scales whereas the hierarchical model performed worse than both. However, using an informed prior from the hierarchical model produces similar errors to using a subject-specific model with large amounts of training data (p<0.1 per subject). The results provide evidence that hierarchical modeling is a promising technique for generalized prediction energy expenditure prediction across a target population in a clinical setting.

Activity recognition using a single accelerometer placed at the wrist or ankle

PURPOSE

Large physical activity surveillance projects such as the UK Biobank and NHANES are using wrist-worn accelerometer-based activity monitors that collect raw data. The goal is to increase wear time by asking subjects to wear the monitors on the wrist instead of the hip, and then to use information in the raw signal to improve activity type and intensity estimation. The purposes of this work was to obtain an algorithm to process wrist and ankle raw data and to classify behavior into four broad activity classes: ambulation, cycling, sedentary, and other activities.

METHODS

Participants (N = 33) wearing accelerometers on the wrist and ankle performed 26 daily activities. The accelerometer data were collected, cleaned, and preprocessed to extract features that characterize 2-, 4-, and 12.8-s data windows. Feature vectors encoding information about frequency and intensity of motion extracted from analysis of the raw signal were used with a support vector machine classifier to identify a subject's activity. Results were compared with categories classified by a human observer. Algorithms were validated using a leave-one-subject-out strategy. The computational complexity of each processing step was also evaluated.

RESULTS

With 12.8-s windows, the proposed strategy showed high classification accuracies for ankle data (95.0%) that decreased to 84.7% for wrist data. Shorter (4 s) windows only minimally decreased performances of the algorithm on the wrist to 84.2%.

CONCLUSIONS

A classification algorithm using 13 features shows good classification into the four classes given the complexity of the activities in the original data set. The algorithm is computationally efficient and could be implemented in real time on mobile devices with only 4-s latency.

Designing a robust activity recognition framework for health and exergaming using wearable sensors

Detecting human activity independent of intensity is essential in many applications, primarily in calculating metabolic equivalent rates and extracting human context awareness. Many classifiers that train on an activity at a subset of intensity levels fail to recognize the same activity at other intensity levels. This demonstrates weakness in the underlying classification method. Training a classifier for an activity at every intensity level is also not practical. In this paper, we tackle a novel intensity-independent activity recognition problem where the class labels exhibit large variability, the data are of high dimensionality, and clustering algorithms are necessary. We propose a new robust stochastic approximation framework for enhanced classification of such data. Experiments are reported using two clustering techniques, K-Means and Gaussian Mixture Models. The stochastic approximation algorithm consistently outperforms other well-known classification schemes which validate the use of our proposed clustered data representation. We verify the motivation of our framework in two applications that benefit from intensity-independent activity recognition. The first application shows how our framework can be used to enhance energy expenditure calculations. The second application is a novel exergaming environment aimed at using games to reward physical activity performed throughout the day, to encourage a healthy lifestyle.

Body weight-normalized Energy Expenditure estimation using combined activity and allometric scaling clustering

Wearable sensors have great potential for accurate estimation of Energy Expenditure (EE) in daily life. Advances in wearable technology (miniaturization, lower costs), and machine learning techniques as well as recently developed self-monitoring movements, such as the Quantified Self, are facilitating mass adoption. However, EE estimations are affected by a person's body weight (BW). BW is a confounding variable preventing meaningful individual and group comparisons. In this paper we present a machine learning approach for BW normalization and activities clustering. In our approach to activity-specific EE modeling, we adopt a genetic algorithm-based clustering scheme, not only based on accelerometer (ACC) features, but also on allometric coefficients derived from 19 subjects performing a wide set of lifestyle and gym activities. We show that our approach supports making comparisons between individuals performing the same activities independently of BW, while maintaining accuracy in the EE estimate.

ReadySteady: app for accelerometer-based activity monitoring and wellness-motivation feedback system for older adults

Increased physical activity and exercise have been found to reduce falls and decrease mortality and age-related morbidity in older adults. However, a large percentage of this population fail to achieve the necessary levels of activity needed to support health living. In this work, we present a mobile app developed on the iOS platform that monitors activity levels using accelerometry. The data captured by the sensor is utilized to provide real-time motivational feedback to enable reinforcement of positive behaviors in older adults. Pilot experiments (conducted with younger adults) performed to assess validity of activity measurement showed that system accurately measures sedentary, light, moderate and vigorous activities in a controlled lab setting. Pilot tests (conducted with older adults) in the user setting showed that while the app is adept at capturing gross body activity (such as sitting, walking and jogging), additional sensors may be required to capture activities involving the extremities.

Classification accuracies of physical activities using smartphone motion sensors

BACKGROUND

Over the past few years, the world has witnessed an unprecedented growth in smartphone use. With sensors such as accelerometers and gyroscopes on board, smartphones have the potential to enhance our understanding of health behavior, in particular physical activity or the lack thereof. However, reliable and valid activity measurement using only a smartphone in situ has not been realized.

OBJECTIVE

To examine the validity of the iPod Touch (Apple, Inc.) and particularly to understand the value of using gyroscopes for classifying types of physical activity, with the goal of creating a measurement and feedback system that easily integrates into individuals' daily living.

METHODS

We collected accelerometer and gyroscope data for 16 participants on 13 activities with an iPod Touch, a device that has essentially the same sensors and computing platform as an iPhone. The 13 activities were sitting, walking, jogging, and going upstairs and downstairs at different paces. We extracted time and frequency features, including mean and variance of acceleration and gyroscope on each axis, vector magnitude of acceleration, and fast Fourier transform magnitude for each axis of acceleration. Different classifiers were compared using the Waikato Environment for Knowledge Analysis (WEKA) toolkit, including C4.5 (J48) decision tree, multilayer perception, naive Bayes, logistic, k-nearest neighbor (kNN), and meta-algorithms such as boosting and bagging. The 10-fold cross-validation protocol was used.

RESULTS

Overall, the kNN classifier achieved the best accuracies: 52.3%-79.4% for up and down stair walking, 91.7% for jogging, 90.1%-94.1% for walking on a level ground, and 100% for sitting. A 2-second sliding window size with a 1-second overlap worked the best. Adding gyroscope measurements proved to be more beneficial than relying solely on accelerometer readings for all activities (with improvement ranging from 3.1% to 13.4%).

CONCLUSIONS

Common categories of physical activity and sedentary behavior (walking, jogging, and sitting) can be recognized with high accuracies using both the accelerometer and gyroscope onboard the iPod touch or iPhone. This suggests the potential of developing just-in-time classification and feedback tools on smartphones.

Bringing the laboratory and clinic to the community: mobile technologies for health promotion and disease prevention

Health-related information collected in psychological laboratories may not be representative of people's everyday health. For at least 70 years, there has been a call for methods that sample experiences from everyday environments and circumstances. New technologies, including cell phones, sensors, and monitors, now make it possible to collect information outside of the laboratory in environments representative of everyday life. We review the role of mobile technologies in the assessment of health-related behaviors, physiological responses, and self-reports. Ecological momentary assessment offers a wide range of new opportunities for ambulatory assessment and evaluation. The value of mobile technologies for interventions to improve health is less well established. Among 21 randomized clinical trials evaluating interventions that used mobile technologies, more than half failed to document significant improvements on health outcomes or health risk factors. Theoretical and practical issues for future research are discussed.

Predicting Adult Pulmonary Ventilation Volume and Wearing Compliance by On-Board Accelerometry During Personal Level Exposure Assessments

BACKGROUND

Metabolic functions typically increase with human activity, but optimal methods to characterize activity levels for real-time predictions of ventilation volume (l/min) during exposure assessments have not been available. Could tiny, triaxial accelerometers be incorporated into personal level monitors to define periods of acceptable wearing compliance, and allow the exposures (μg/m3) to be extended to potential doses in μg/min/kg of body weight?

OBJECTIVES

In a pilot effort, we tested: 1) whether appropriately-processed accelerometer data could be utilized to predict compliance and in linear regressions to predict ventilation volumes in real time as an on-board component of personal level exposure sensor systems, and 2) whether locating the exposure monitors on the chest in the breathing zone, provided comparable accelerometric data to other locations more typically utilized (waist, thigh, wrist, etc.).

METHODS

Prototype exposure monitors from RTI International and Columbia University were worn on the chest by a pilot cohort of adults while conducting an array of scripted activities (all <10 METS), spanning common recumbent, sedentary, and ambulatory activity categories. Referee Wocket accelerometers that were placed at various body locations allowed comparison with the chest-located exposure sensor accelerometers. An Oxycon Mobile mask was used to measure oral-nasal ventilation volumes in-situ. For the subset of participants with complete data (n= 22), linear regressions were constructed (processed accelerometric variable versus ventilation rate) for each participant and exposure monitor type, and Pearson correlations computed to compare across scenarios.

RESULTS

Triaxial accelerometer data were demonstrated to be adequately sensitive indicators for predicting exposure monitor wearing compliance. Strong linear correlations (R values from 0.77 to 0.99) were observed for all participants for both exposure sensor accelerometer variables against ventilation volume for recumbent, sedentary, and ambulatory activities with MET values ~<6. The RTI monitors mean R value of 0.91 was slightly higher than the Columbia monitors mean of 0.86 due to utilizing a 20 Hz data rate instead of a slower 1 Hz rate. A nominal mean regression slope was computed for the RTI system across participants and showed a modest RSD of +/-36.6%. Comparison of the correlation values of the exposure monitors with the Wocket accelerometers at various body locations showed statistically identical regressions for all sensors at alternate hip, ankle, upper arm, thigh, and pocket locations, but not for the Wocket accelerometer located at the dominant-side wrist location (R=0.57; p=0.016).

CONCLUSIONS

Even with a modest number of adult volunteers, the consistency and linearity of regression slopes for all subjects were very good with excellent within-person Pearson correlations for the accelerometer versus ventilation volume data. Computing accelerometric standard deviations allowed good sensitivity for compliance assessments even for sedentary activities. These pilot findings supported the hypothesis that a common linear regression is likely to be usable for a wider range of adults to predict ventilation volumes from accelerometry data over a range of low to moderate energy level activities. The predicted volumes would then allow real-time estimates of potential dose, enabling more robust panel studies. The poorer correlation in predicting ventilation rate for an accelerometer located on the wrist suggested that this location should not be considered for predictions of ventilation volume.

Sedentary behavior: target for change, challenge to assess

Sedentary behavior is not a new topic, but trying to examine the direct links between sedentary behavior and health outcomes, independent of time spent in moderate- and vigorous-intensity physical activity, is a relatively new addition to the relationships between physical activity and health. Defining sedentary behavior as a risk factor and target for intervention opens up novel avenues for disease prevention and health promotion. The relationship between sedentary behavior and obesity is complex and not well understood, but the increased risk of disease due to sedentary behavior may be even greater in obese patients. Objective measurement of sedentary behavior is an important link in being able to understand the real effects of being sedentary, and a few measurement devices are described. Interventions targeting sedentary behavior should reduce total sedentary time, break long bouts of sitting with intermittent activity and encourage light-intensity activity throughout the day. New technologies can both measure and deliver an intervention aimed at reducing sitting time, the most common category of sedentary behavior. An optimal activity profile will include minimal amounts of sedentary behavior, in addition to regular physical activity and healthy sleep patterns.

Design of a wearable physical activity monitoring system using mobile phones and accelerometers

This paper describes the motivation for, and overarching design of, an open-source hardware and software system to enable population-scale, longitudinal measurement of physical activity and sedentary behavior using common mobile phones. The "Wockets" data collection system permits researchers to collect raw motion data from participants who wear multiple small, comfortable sensors for 24 hours per day, including during sleep, and monitor data collection remotely.

Determining energy expenditure from treadmill walking using hip-worn inertial sensors: an experimental study

We describe an experimental study to estimate energy expenditure during treadmill walking using a single hip-mounted inertial sensor (triaxial accelerometer and triaxial gyroscope). Typical physical-activity characterization using commercial monitors use proprietary counts that do not have a physically interpretable meaning. This paper emphasizes the role of probabilistic techniques in conjunction with inertial data modeling to accurately predict energy expenditure for steady-state treadmill walking. We represent the cyclic nature of walking with a Fourier transform and show how to map this representation to energy expenditure (VO(2), mL/min) using three regression techniques. A comparative analysis of the accuracy of sensor streams in predicting energy expenditure reveals that using triaxial information leads to more accurate energy-expenditure prediction compared to only using one axis. Combining accelerometer and gyroscope information leads to improved accuracy compared to using either sensor alone. Nonlinear regression methods showed better prediction accuracy compared to linear methods but required an order of higher magnitude run time.

Evaluation of artificial neural network algorithms for predicting METs and activity type from accelerometer data: validation on an independent sample

Previous work from our laboratory provided a "proof of concept" for use of artificial neural networks (nnets) to estimate metabolic equivalents (METs) and identify activity type from accelerometer data (Staudenmayer J, Pober D, Crouter S, Bassett D, Freedson P, J Appl Physiol 107: 1330-1307, 2009). The purpose of this study was to develop new nnets based on a larger, more diverse, training data set and apply these nnet prediction models to an independent sample to evaluate the robustness and flexibility of this machine-learning modeling technique. The nnet training data set (University of Massachusetts) included 277 participants who each completed 11 activities. The independent validation sample (n = 65) (University of Tennessee) completed one of three activity routines. Criterion measures were 1) measured METs assessed using open-circuit indirect calorimetry; and 2) observed activity to identify activity type. The nnet input variables included five accelerometer count distribution features and the lag-1 autocorrelation. The bias and root mean square errors for the nnet MET trained on University of Massachusetts and applied to University of Tennessee were +0.32 and 1.90 METs, respectively. Seventy-seven percent of the activities were correctly classified as sedentary/light, moderate, or vigorous intensity. For activity type, household and locomotion activities were correctly classified by the nnet activity type 98.1 and 89.5% of the time, respectively, and sport was correctly classified 23.7% of the time. Use of this machine-learning technique operates reasonably well when applied to an independent sample. We propose the creation of an open-access activity dictionary, including accelerometer data from a broad array of activities, leading to further improvements in prediction accuracy for METs, activity intensity, and activity type.