Dataset Information

Identifying Free-Living Physical Activities Using Lab-Based Models with Wearable Accelerometers.

ABSTRACT: The purpose of this study was to classify, and model various physical activities performed by a diverse group of participants in a supervised lab-based protocol and utilize the model to identify physical activity in a free-living setting. Wrist-worn accelerometer data were collected from ( N = 152 ) adult participants; age 18?64 years, and processed the data to identify and model unique physical activities performed by the participants in controlled settings. The Gaussian mixture model (GMM) and the hidden Markov model (HMM) algorithms were used to model the physical activities with time and frequency-based accelerometer features. An overall model accuracy of 92.7% and 94.7% were achieved to classify 24 physical activities using GMM and HMM, respectively. The most accurate model was then used to identify physical activities performed by 20 participants, each recorded for two free-living sessions of approximately six hours each. The free-living activity intensities were estimated with 80% accuracy and showed the dominance of stationary and light intensity activities in 36 out of 40 recorded sessions. This work proposes a novel activity recognition process to identify unsupervised free-living activities using lab-based classification models. In summary, this study contributes to the use of wearable sensors to identify physical activities and estimate energy expenditure in free-living settings.

SUBMITTER: Dutta A

PROVIDER: S-EPMC6263387 | biostudies-literature | 2018 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Identifying Free-Living Physical Activities Using Lab-Based Models with Wearable Accelerometers.

Dutta Arindam A Ma Owen O Toledo Meynard M Pregonero Alberto Florez AF Ainsworth Barbara E BE Buman Matthew P MP Bliss Daniel W DW

Sensors (Basel, Switzerland) 20181112 11

The purpose of this study was to classify, and model various physical activities performed by a diverse group of participants in a supervised lab-based protocol and utilize the model to identify physical activity in a free-living setting. Wrist-worn accelerometer data were collected from ( N = 152 ) adult participants; age 18⁻64 years, and processed the data to identify and model unique physical activities performed by the participants in controlled settings. The Gaussian mixture model (GMM) and ...[more]

PMID: 30424512

Similar Datasets

Project description:Physical activity levels are related through algorithms to the energetic demand, with no information regarding the integrity of the multiple physiological systems involved in the energetic supply. Longitudinal analysis of the oxygen uptake (V?o2) by wearable sensors in realistic settings might permit development of a practical tool for the study of the longitudinal aerobic system dynamics (i.e., V?o2 kinetics). This study evaluated aerobic system dynamics based on predicted V?o2 data obtained from wearable sensors during unsupervised activities of daily living (?ADL). Thirteen healthy men performed a laboratory-controlled moderate exercise protocol and were monitored for ?6 h/day for 4 days (?ADL data). Variables derived from hip accelerometer (ACCHIP), heart rate monitor, and respiratory bands during ?ADL were extracted and processed by a validated random forest regression model to predict V?o2. The aerobic system analysis was based on the frequency-domain analysis of ACCHIP and predicted V?o2 data obtained during ?ADL. Optimal samples for frequency domain analysis (constrained to ?0.01 Hz) were selected when ACCHIP was higher than 0.05 g at a given frequency (i.e., participants were active). The temporal characteristics of predicted V?o2 data during ?ADL correlated with the temporal characteristics of measured V?o2 data during laboratory-controlled protocol ([Formula: see text]?=?0.82, P < 0.001, n?=?13). In conclusion, aerobic system dynamics can be investigated during unsupervised activities of daily living by wearable sensors. Although speculative, these algorithms have the potential to be incorporated into wearable systems for early detection of changes in health status in realistic environments by detecting changes in aerobic response dynamics. NEW & NOTEWORTHY The early detection of subclinical aerobic system impairments might be indicative of impaired physiological reserves that impact the capacity for physical activity. This study is the first to use wearable sensors in unsupervised activities of daily living in combination with novel machine learning algorithms to investigate the aerobic system dynamics with the potential to contribute to models of functional health status and guide future individualized health care in the normal population.

Project description:ContextAnterior cruciate ligament reconstruction (ACLR) and gait speed are risk factors for developing knee osteoarthritis (OA). Measuring minute-level cadence during free-living activities may aid in identifying individuals at elevated risk of developing slow habitual gait speed and, in the long term, OA.ObjectiveTo assess differences in peak 1-minute cadence and weekly time in different cadence intensities between individuals with and without ACLR.DesignCross-sectional study.SettingShort-term, free-living conditions.Patients or other participantsA total of 57 participants with ACLR (34 women, 23 men; age = 20.9 ± 3.2 years, time since surgery = 28.7 ± 17.7 months) and 42 healthy control participants (22 women, 20 men; age = 20.7 ± 1.7 years).Main outcome measure(s)Each participant wore a physical activity monitor for 7 days. Data were collected at 30 Hz, processed in 60-second epochs, and included in the analyses if the activity monitor was worn for at least 10 hours per day over 4 days. Mean daily steps, peak 1-minute cadence, and weekly minutes spent at 60 to 79 (slow walking), 80 to 99 (medium walking), 100 to 119 (brisk walking), ≥100 (moderate- to vigorous-intensity ambulation), and ≥130 (vigorous-intensity ambulation) steps per minute were calculated. One-way analyses of covariance were conducted to determine differences between groups, controlling for height and activity-monitor wear time.ResultsThose with ACLR took fewer daily steps (8422 ± 2663 versus 10 033 ± 3046 steps; P = .005) and spent fewer weekly minutes in moderate- to vigorous-intensity cadence (175.8 ± 116.5 minutes versus 218.5 ± 137.1 minutes; P = .048) than participants without ACLR. We observed no differences in minutes spent at slow (ACLR = 77.4 ± 40.5 minutes versus control = 83.9 ± 34.3 minutes; P = .88), medium (ACLR = 71.6 ± 40.2 minutes versus control = 82.9 ± 46.8 minutes; P = .56), brisk (ACLR = 115.3 ± 70.3 minutes versus control = 138.3 ± 73.3 minutes; P = .18), or vigorous-intensity (ACLR = 24.3 ± 36.5 minutes versus control = 38.1 ± 60.9 minutes; P = .10) cadences per week.ConclusionsParticipants with ACLR walked approximately 40 fewer minutes per week in moderate- to vigorous-intensity cadence than participants without ACLR. Increasing the time spent at cadence ≥100 steps per minute and overall volume of physical activity may be useful as interventional targets to help reduce the risk of early development of OA after ACLR.

Project description:Background: Physical activity (PA) is essential in stroke rehabilitation of hemiparetic patients to avoid health risks, and moderate to vigorous PA could promote patients' recovery. However, PA assessments are limited to clinical environments. Little is known about PA in unguided free-living. Wearable sensors could reveal patients' PA during rehabilitation, and day-long long-term measurements over several weeks might reveal recovery trends of affected and less-affected body sides. Methods: We investigated PA in an observation study during outpatient rehabilitation in a day-care center. PA of affected and less-affected body sides, including upper and lower limbs were derived using wearable motion sensors. In this analysis we focused on PA during free-living and clinician guided therapies, and investigated differences between body-sides. Linear regressions were used to estimate metabolic equivalents for each limb at comparable scale. Non-parametric statistics were derived to quantify PA differences between body sides. Results: We analyzed 102 full-day movement data recordings from eleven hemiparetic patients during individual rehabilitation periods up to 79 days. The comparison between free-living and clinician guided therapy showed on average 16.1 % higher PA in the affected arm during therapy and 5.3 % higher PA in the affected leg during therapy. Average differences between free-living and therapy in the less-affected side were below 4.5 %. Conclusion: We analyzed PA of patients with a hemiparesis in two distinct rehabilitation settings, including free-living and clinician guided therapies over several weeks and compared MET values of affected and less-affected body sides. In particular, we investigated PA using individual regression models for each limb. We demonstrated that wearable motion sensors provide insights in patient's PA during rehabilitation. Although, no clear PA trends were found, our analysis showed patients' tendency to sedentary behavior, confirming previous lab study results. Our PA analysis approach could be used beyond clinical rehabilitation to devise personalized patient and limb-specific exercise recommendations in future remote rehabilitation.

Project description:Introduction: Understanding determinants of community mobility disability is critical for developing interventions aimed at preventing or delaying disability in older adults. In an effort to understand these determinants, capturing and measuring community mobility has become a key factor. The objectives of this paper are to present and illustrate the signal processing workflow and outcomes that can be extracted from an activity and community mobility measurement approach based on GPS and accelerometer sensor data and 2) to explore the construct validity of the proposed measurement approach using data collected from healthy older adults in free-living conditions. Methods: Personal, functional impairment and environmental variables were obtained by self-report questionnaires in 75 healthy community-living older adults (mean age = 66 ± 7 years old) living on the island of Montreal, QC, Canada. Participants wore, for 14 days during waking hours on the hip, a data logger incorporating a GPS receiver with a 3-axis accelerometer. Time at home ratio (THR), Trips out (TO), Destinations (D), Maximal distance of destinations (MDD), Active time ratio (ATR), Steps (S), Distance in a vehicle (DV), Time in a vehicle (TV), Distance on foot (DF), Time on foot (TF), Ellipse area (EA), and Ellipse maximum distance (EMD) were extracted from the recordings. Results: After applying quality control criteria, the original data set was reduced from 75 to 54 participants (28% attrition). Results from the remaining sample show that under free-living conditions in healthy older adults, location, activity and community mobility outcomes vary across individuals and certain personal variables (age, income, living situation, professional status, vehicle access) have potential mitigating effects on these outcomes. There was a significant (yet small) relationship (rho < 0.40) between self-reported life space and MDD, DV, EA, and EMD. Conclusion: Wearability and usability of the devices used to capture free-living community mobility impact participant compliance and the quality of the data. The construct validity of the proposed approach appears promising but requires further studies directed at populations with mobility impairments.

Project description:PurposeTo adapt and validate a previously developed decision tree for youth to identify bedrest for use in preschool children.MethodsParents of healthy preschool (3-6-year-old) children (n = 610; 294 males) were asked to help them to wear an accelerometer for 7 to 10 days and 24 hours/day on their waist. Children with ≥3 nights of valid recordings were randomly allocated to the development (n = 200) and validation (n = 200) groups. Wear periods from accelerometer recordings were identified minute-by-minute as bedrest or wake using visual identification by two independent raters. To automate visual identification, chosen decision tree (DT) parameters (block length, threshold, bedrest-start trigger, and bedrest-end trigger) were optimized in the development group using a Nelder-Mead simplex optimization method, which maximized the accuracy of DT-identified bedrest in 1-min epochs against synchronized visually identified bedrest (n = 4,730,734). DT's performance with optimized parameters was compared with the visual identification, commonly used Sadeh's sleep detection algorithm, DT for youth (10-18-years-old), and parental survey of sleep duration in the validation group.ResultsOn average, children wore an accelerometer for 8.3 days and 20.8 hours/day. Comparing the DT-identified bedrest with visual identification in the validation group yielded sensitivity = 0.941, specificity = 0.974, and accuracy = 0.956. The optimal block length was 36 min, the threshold 230 counts/min, the bedrest-start trigger 305 counts/min, and the bedrest-end trigger 1,129 counts/min. In the validation group, DT identified bedrest with greater accuracy than Sadeh's algorithm (0.956 and 0.902) and DT for youth (0.956 and 0.861) (both P<0.001). Both DT (564±77 min/day) and Sadeh's algorithm (604±80 min/day) identified significantly less bedrest/sleep than parental survey (650±81 min/day) (both P<0.001).ConclusionsThe DT-based algorithm initially developed for youth was adapted for preschool children to identify time spent in bedrest with high accuracy. The DT is available as a package for the R open-source software environment ("PhysActBedRest").

Dataset Information

Identifying Free-Living Physical Activities Using Lab-Based Models with Wearable Accelerometers.

Publications

Identifying Free-Living Physical Activities Using Lab-Based Models with Wearable Accelerometers.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets