A balance between the feasibility and validity of measures is an important consideration for physical activity research - particularly in school-based research with youth. The present study extends previously tested calibration methods to develop and test new equations for an online version of the Youth Activity Profile (YAP) tool, a self-report tool designed for school applications. Data were collected across different regions and seasons to develop more robust, generalizable equations. The study involved a total of 717 youth from 33 schools (374 elementary (ages 9-11), 224 middle (ages 11-14), and 119 high school (ages 14-18)) in two different states in the U.S. Participants wore a Sensewear monitor for a full week and then completed the online YAP at school to report physical activity (PA) and sedentary behaviors (SB) in school and at home. Accelerometer data were processed using an R-based segmentation program to compute PA and SB levels. Quantile regression models were used with half of the sample to develop item-specific YAP calibration equations and these were cross validated with the remaining half of the sample. Computed values of Mean Absolute Percent Error (MAPE) ranged from 15-25% with slightly lower error observed for the middle school sample. The new equations had improved precision compared to the previous versions when tested on the same sample. The online version of the YAP provides an efficient and effective way to capture school level estimates of PA and SB in youth.
Background: Machine learning has been used for classification of physical behavior bouts from hip-worn accelerometers; however, this research has been limited due to the challenges of directly observing and coding human behavior "in the wild." Deep learning algorithms, such as convolutional neural networks (CNNs), may offer better representation of data than other machine learning algorithms without the need for engineered features and may be better suited to dealing with free-living data. The purpose of this study was to develop a modeling pipeline for evaluation of a CNN model on a free-living data set and compare CNN inputs and results with the commonly used machine learning random forest and logistic regression algorithms.
Method: Twenty-eight free-living women wore an ActiGraph GT3X+accelerometer on their right hip for 7 days. A concurrently worn thigh-mounted activPAL device captured ground truth activity labels. The authors evaluated logistic regression, random forest, and CNN models for classifying sitting, standing, and stepping bouts. The authors also assessed the benefit of performing feature engineering for this task.
Results: The CNN classifier performed best (average balanced accuracy for bout classification of sitting, standing, and stepping was 84%) compared with the other methods (56% for logistic regression and 76% for random forest), even without performing any feature engineering.
Conclusion: Using the recent advancements in deep neural networks, the authors showed that a CNN model can outperform other methods even without feature engineering. This has important implications for both the model's ability to deal with the complexity of free-living data and its potential transferability to new populations.
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：