Background: This paper examined whether the criterion validity of step count (SC), energy expenditure (EE), and heart rate (HR) varied across studies depending on the average age, body mass index (BMI), and predominant gender of participants. Methods: Data from 1536 studies examining the validity of various wearable devices were used. Separate multilevel regression models examined the associations among age, gender, and BMI with device criterion validity assessed using mean absolute percent error (MAPE) at the study level. Results: MAPE values were reported in 970 studies for SC, 328 for EE, and 238 for HR, respectively. There were several significant differences in MAPE between age, gender, and BMI categories for SC, EE, and HR. SC MAPE was significantly different for older adults compared with adults. Compared with studies among normal-weight populations, MAPE was greater among studies with overweight samples for SC, HR, and EE. Comparing studies with more women than men, MAPE was significantly greater for EE and HR. Conclusions: There are important differences in the criterion validity of commercial wearable devices across studies of varying ages, BMIs, and genders. Few studies have examined differences in error between different age groups, particularly for EE and HR.
Many studies have explored divergent deep neural networks in human activity recognition (HAR) using a single accelerometer sensor. Multiple types of deep neural networks, such as convolutional neural networks (CNN), long short-term memory (LSTM), or their hybridization (CNN-LSTM), have been implemented. However, the sensor orientation problem poses challenges in HAR, and the length of windows as inputs for the deep neural networks has mostly been adopted arbitrarily. This paper explores the effect of window lengths with orientation invariant heuristic features on the performance of 1D-CNN-LSTM in recognizing six human activities; sitting, lying, walking and running at three different speeds using data from an accelerometer sensor encapsulated into a smartphone. Forty-two participants performed the six mentioned activities by keeping smartphones in their pants pockets with arbitrary orientation. We conducted an inter-participant evaluation using 1D-CNN-LSTM architecture. We found that the average accuracy of the classifier was saturated to 80 ± 8.07% for window lengths greater than 65 using only four selected simple orientation invariant heuristic features. In addition, precision, recall and F1-measure in recognizing stationary activities such as sitting and lying decreased with increment of window length, whereas we encountered an increment in recognizing the non-stationary activities.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.