Suneeta Godbole

Using the SenseCam to Improve Classifications of Sedentary Behavior in Free-Living Settings

BACKGROUND Studies have shown relationships between important health outcomes and sedentary behavior, independent of physical activity. There are known errors in tools employed to assess sedentary behavior. Studies of accelerometers have been limited to laboratory environments. PURPOSE To assess a broad range of sedentary behaviors in free-living adults using accelerometers and a Microsoft SenseCam that can provide an objective observation of sedentary behaviors through first person-view images. METHODS Participants were 40 university employees who wore a SenseCam and Actigraph accelerometer for 3-5 days. Images were coded for sitting and standing posture and 12 activity types. Data were merged and aggregated to a 60-second epoch. Accelerometer counts per minute (cpm) of <100 were compared with coded behaviors. Sensitivity and specificity analyses were performed. Data were collected in June and July 2011 and analyzed in April 2012. RESULTS TV viewing, other screen use, and administrative activities were correctly classified by the 100-cpm cutpoint. However, standing behaviors also fell under this threshold, and driving behaviors exceeded it. Multiple behaviors occurred simultaneously. A nearly 30-minute per day difference was found in sedentary behavior estimates based on the accelerometer versus the SenseCam. CONCLUSIONS Researchers should be aware of the strengths and weaknesses of the 100-cpm accelerometer cutpoint for identifying sedentary behavior. The SenseCam may be a useful tool in free-living conditions to better understand health behaviors such as sitting.

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.

The Relationship Between Outdoor Activity and Health in Older Adults Using GPS

Physical activity (PA) provides health benefits in older adults. Research suggests that exposure to nature and time spent outdoors may also have effects on health. Older adults are the least active segment of our population, and are likely to spend less time outdoors than other age groups. The relationship between time spent in PA, outdoor time, and various health outcomes was assessed for 117 older adults living in retirement communities. Participants wore an accelerometer and GPS device for 7 days. They also completed assessments of physical, cognitive, and emotional functioning. Analyses of variance were employed with a main and interaction effect tested for ±30 min PA and outdoor time. Significant differences were found for those who spent >30 min in PA or outdoors for depressive symptoms, fear of falling, and self-reported functioning. Time to complete a 400 m walk was significantly different by PA time only. QoL and cognitive functioning scores were not significantly different. The interactions were also not significant. This study is one of the first to demonstrate the feasibility of using accelerometer and GPS data concurrently to assess PA location in older adults. Future analyses will shed light on potential causal relationships and could inform guidelines for outdoor activity.

Identifying Active Travel Behaviors in Challenging Environments Using GPS, Accelerometers, and Machine Learning Algorithms

Background: Active travel is an important area in physical activity research, but objective measurement of active travel is still difficult. Automated methods to measure travel behaviors will improve research in this area. In this paper, we present a supervised machine learning method for transportation mode prediction from global positioning system (GPS) and accelerometer data. Methods: We collected a dataset of about 150 h of GPS and accelerometer data from two research assistants following a protocol of prescribed trips consisting of five activities: bicycling, riding in a vehicle, walking, sitting, and standing. We extracted 49 features from 1-min windows of this data. We compared the performance of several machine learning algorithms and chose a random forest algorithm to classify the transportation mode. We used a moving average output filter to smooth the output predictions over time. Results: The random forest algorithm achieved 89.8% cross-validated accuracy on this dataset. Adding the moving average filter to smooth output predictions increased the cross-validated accuracy to 91.9%. Conclusion: Machine learning methods are a viable approach for automating measurement of active travel, particularly for measuring travel activities that traditional accelerometer data processing methods misclassify, such as bicycling and vehicle travel.

Validity of Palms Gps Scoring of Active and Passive Travel Compared with Sensecam

PURPOSE The objective of this study is to assess validity of the personal activity location measurement system (PALMS) for deriving time spent walking/running, bicycling, and in vehicle, using SenseCam (Microsoft, Redmond, WA) as the comparison. METHODS Forty adult cyclists wore a Qstarz BT-Q1000XT GPS data logger (Qstarz International Co., Taipei, Taiwan) and SenseCam (camera worn around the neck capturing multiple images every minute) for a mean time of 4 d. PALMS used distance and speed between global positioning system (GPS) points to classify whether each minute was part of a trip (yes/no), and if so, the trip mode (walking/running, bicycling, or in vehicle). SenseCam images were annotated to create the same classifications (i.e., trip yes/no and mode). Contingency tables (2 × 2) and confusion matrices were calculated at the minute level for PALMS versus SenseCam classifications. Mixed-effects linear regression models estimated agreement (mean differences and intraclass correlation coefficients) between PALMS and SenseCam with regard to minutes/day in each mode. RESULTS Minute-level sensitivity, specificity, and negative predictive value were ≥88%, and positive predictive value was ≥75% for non-mode-specific trip detection. Seventy-two percent to 80% of outdoor walking/running minutes, 73% of bicycling minutes, and 74%-76% of in-vehicle minutes were correctly classified by PALMS. For minutes per day, PALMS had a mean bias (i.e., amount of over or under estimation) of 2.4-3.1 min (11%-15%) for walking/running, 2.3-2.9 min (7%-9%) for bicycling, and 4.3-5 min (15%-17%) for vehicle time. Intraclass correlation coefficients were ≥0.80 for all modes. CONCLUSIONS PALMS has validity for processing GPS data to objectively measure time spent walking/running, bicycling, and in vehicle in population studies. Assessing travel patterns is one of many valuable applications of GPS in physical activity research that can improve our understanding of the determinants and health outcomes of active transportation as well as its effect on physical activity.

Hip and Wrist Accelerometer Algorithms for Free-living Behavior Classification

PURPOSE Accelerometers are a valuable tool for objective measurement of physical activity (PA). Wrist-worn devices may improve compliance over standard hip placement, but more research is needed to evaluate their validity for measuring PA in free-living settings. Traditional cut-point methods for accelerometers can be inaccurate and need testing in free living with wrist-worn devices. In this study, we developed and tested the performance of machine learning (ML) algorithms for classifying PA types from both hip and wrist accelerometer data. METHODS Forty overweight or obese women (mean age = 55.2 ± 15.3 yr; BMI = 32.0 ± 3.7) wore two ActiGraph GT3X+ accelerometers (right hip, nondominant wrist; ActiGraph, Pensacola, FL) for seven free-living days. Wearable cameras captured ground truth activity labels. A classifier consisting of a random forest and hidden Markov model classified the accelerometer data into four activities (sitting, standing, walking/running, and riding in a vehicle). Free-living wrist and hip ML classifiers were compared with each other, with traditional accelerometer cut points, and with an algorithm developed in a laboratory setting. RESULTS The ML classifier obtained average values of 89.4% and 84.6% balanced accuracy over the four activities using the hip and wrist accelerometer, respectively. In our data set with average values of 28.4 min of walking or running per day, the ML classifier predicted average values of 28.5 and 24.5 min of walking or running using the hip and wrist accelerometer, respectively. Intensity-based cut points and the laboratory algorithm significantly underestimated walking minutes. CONCLUSIONS Our results demonstrate the superior performance of our PA-type classification algorithm, particularly in comparison with traditional cut points. Although the hip algorithm performed better, additional compliance achieved with wrist devices might justify using a slightly lower performing algorithm.

Multi-sensor physical activity recognition in free-living

Physical activity monitoring in free-living populations has many applications for public health research, weight-loss interventions, context-aware recommendation systems and assistive technologies. We present a system for physical activity recognition that is learned from a free-living dataset of 40 women who wore multiple sensors for seven days. The multi-level classification system first learns low-level codebook representations for each sensor and uses a random forest classifier to produce minute-level probabilities for each activity class. Then a higher-level HMM layer learns patterns of transitions and durations of activities over time to smooth the minute-level predictions.

Objective Assessment of Physical Activity: Classifiers for Public Health.

PURPOSE Walking for health is recommended by health agencies, partly based on epidemiological studies of self-reported behaviors. Accelerometers are now replacing survey data, but it is not clear that intensity-based cut points reflect the behaviors previously reported. New computational techniques can help classify raw accelerometer data into behaviors meaningful for public health. METHODS Five hundred twenty days of triaxial 30-Hz accelerometer data from three studies (n = 78) were employed as training data. Study 1 included prescribed activities completed in natural settings. The other two studies included multiple days of free-living data with SenseCam-annotated ground truth. The two populations in the free-living data sets were demographically and physical different. Random forest classifiers were trained on each data set, and the classification accuracy on the training data set and that applied to the other available data sets were assessed. Accelerometer cut points were also compared with the ground truth from the three data sets. RESULTS The random forest classified all behaviors with over 80% accuracy. Classifiers developed on the prescribed data performed with higher accuracy than the free-living data classifier, but these did not perform as well on the free-living data sets. Many of the observed behaviors occurred at different intensities compared with those identified by existing cut points. CONCLUSIONS New machine learning classifiers developed from prescribed activities (study 1) were considerably less accurate when applied to free-living populations or to a functionally different population (studies 2 and 3). These classifiers, developed on free-living data, may have value when applied to large cohort studies with existing hip accelerometer data.

Two-arm randomized pilot intervention trial to decrease sitting time and increase sit-to-stand transitions in working and non-working older adults

Background Excessive sitting has been linked to poor health. It is unknown whether reducing total sitting time or increasing brief sit-to-stand transitions is more beneficial. We conducted a randomized pilot study to assess whether it is feasible for working and non-working older adults to reduce these two different behavioral targets. Methods Thirty adults (15 workers and 15 non-workers) age 50–70 years were randomized to one of two conditions (a 2-hour reduction in daily sitting or accumulating 30 additional brief sit-to-stand transitions per day). Sitting time, standing time, sit-to-stand transitions and stepping were assessed by a thigh worn inclinometer (activPAL). Participants were assessed for 7 days at baseline and followed while the intervention was delivered (2 weeks). Mixed effects regression analyses adjusted for days within participants, device wear time, and employment status. Time by condition interactions were investigated. Results Recruitment, assessments, and intervention delivery were feasible. The ‘reduce sitting’ group reduced their sitting by two hours, the ‘increase sit-to-stand’ group had no change in sitting time (p < .001). The sit-to-stand transition group increased their sit-to-stand transitions, the sitting group did not (p < .001). Conclusions This study was the first to demonstrate the feasibility and preliminary efficacy of specific sedentary behavioral goals. Trial Registration clinicaltrials.gov NCT02544867

Neighborhood preference, walkability and walking in overweight/obese men.

OBJECTIVES To investigate whether self-selection moderated the effects of walkability on walking in overweight and obese men. METHODS 240 overweight and obese men completed measures on importance of walkability when choosing a neighborhood (selection) and preference for walkable features in general (preference). IPAQ measured walking. A walkbility index was derived from geographic information systems (GIS). RESULTS Walkability was associated with walking for transportation (p = .027) and neighborhood selection was associated with walking for transportation (p = .002) and total walking (p = .001). Preference was associated with leisure walking (p = .045) and preference moderated the relationship between walkability and total walking (p = .059). CONCLUSION Walkability and self-selection are both important to walking behavior.