Abstract
Machine-learning (ML) approaches have been repeatedly coupled with raw accelerometry to classify physical activity classes, but the features required to optimize their predictive performance are still unknown. Our aim was to identify appropriate combination of feature subsets and prediction algorithms for activity class prediction from hip-based raw acceleration data. The hip-based raw acceleration data collected from 27 participants was split into training (70 %) and validation (30 %) subsets. A total of 206 time- (TD) and frequencydomain (FD) features were extracted from 6-second non-overlapping windows of the signal. Feature selection was done using seven filter-based, two wrapper-based, and one embedded algorithm, and classification was performed with artificial neural network (ANN), support vector machine (SVM), and random forest (RF). For every combination between the feature selection method and the classifiers, the most appropriate feature subsets were found and used for model training within the training set. These models were then validated with the left-out validation set. The appropriate number of features for the ANN, SVM, and RF ranged from 20 to 45. Overall, the accuracy of all the three classifiers was higher when trained with feature subsets generated using filter-based methods compared with when they were trained with wrapper-based methods (range: 78.1 %-88 % vs. 66 %-83.5 %). TD features that reflect how signals vary around the mean, how they differ with one another, and how much and how often they change were more frequently selected via the feature selection methods. A subset of TD features from raw accelerometry could be sufficient for ML-based activity classification if properly selected from different axes.
Highlights
Accelerometers are the most commonly used wearable sensors in large-scale observational studies for the device-based measurement of daily activities because of their feasibility and applicability [1]
The artificial neural networks (ANN) model performed well with all the algorithmically selected feature sets, but it performed best when it was trained with the feature subset selected according to symmetrical uncertainty using 20 features
The support vector ma chines (SVM) achieved its highest performance when trained with the feature subset selected according to symmetrical uncertainty using 45 features, but had comparable performance when trained with the feature subsets selected based on Information gain and Gain Ratio using 40 and 30 features, respectively
Summary
Accelerometers are the most commonly used wearable sensors in large-scale observational studies for the device-based measurement of daily activities because of their feasibility and applicability [1]. The equations and cut points established based on those approaches have been repeatedly shown to possess limited prediction accuracy across a range of activities commonly performed in daily life (from sedentary behavior to vigorous physical activities) [3]. This has resulted in the emergence of advanced machine learning (ML)–based approaches for activity class prediction from raw accelerometer signals [4]. Few studies have algorithmically found the appropriate subset of features from raw acceleration data prior to developing the model [11,12,13,14,15] with a limited pool of features (
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
