Multimodal Physical Activity Recognition by Fusing Temporal and Cepstral Information

Abstract

A physical activity (PA) recognition algorithm for a wearable wireless sensor network using both ambulatory electrocardiogram (ECG) and accelerometer signals is proposed. First, in the time domain, the cardiac activity mean and the motion artifact noise of the ECG signal are modeled by a Hermite polynomial expansion and principal component analysis, respectively. A set of time domain accelerometer features is also extracted. A support vector machine (SVM) is employed for supervised classification using these time domain features. Second, motivated by their potential for handling convolutional noise, cepstral features extracted from ECG and accelerometer signals based on a frame level analysis are modeled using Gaussian mixture models (GMMs). Third, to reduce the dimension of the tri-axial accelerometer cepstral features which are concatenated and fused at the feature level, heteroscedastic linear discriminant analysis is performed. Finally, to improve the overall recognition performance, fusion of the multi-modal (ECG and accelerometer) and multidomain (time domain SVM and cepstral domain GMM) subsystems at the score level is performed. The classification accuracy ranges from 79.3% to 97.3% for various testing scenarios and outperforms the state-of-the-art single accelerometer based PA recognition system by over 24% relative error reduction on our nine-category PA database.

Innovations in the Use of Interactive Technology to Support Weight Management

“…These technologies and cut-points to determine energy expenditure have been rigorously validated in various populations 12, 13 . Pattern recognition techniques can be developed to recognize specific activities as they occur, such as sitting, standing or walking, using data from sensors and smartphones 14–16 . More recently, popular wearable devices and phone apps for capturing physical activity have also been tested for accuracy 17 , and depending upon the research question and main outcomes, these, too can be used to inform interventions.…”

“…Many off-the-shelf accelerometers and activity trackers do not yet provide streaming data and/or an open application programming interface (API), which allows researchers to access data and harmonize software and hardware. Therefore, some researchers continue to develop and test their own accelerometers 18 , or their own applications that can derive momentary physical activity estimates from the raw accelerometer data obtained by smartphones 14, 19 .…”

Understanding physiological responses to stressors during physical activity

“…First, the physiological system reacts not only to changes in stress, but also to many other factors such as changes in physical or mental conditions; thus, a physiological change does not necessarily imply a stress change [5,9]. It often responds to physical activity demands, physical discomfort, noise, changes in posture, lighting conditions and mental task demand and emotional stress.…”

“…It often responds to physical activity demands, physical discomfort, noise, changes in posture, lighting conditions and mental task demand and emotional stress. In fact, researchers have exploited physiological responses in recognizing physical activity [9]. Second, sensing platforms for detecting physiological signals have limitations [10].…”

RPCA-based detection and quantification of motion artifacts in ECG signals

“…Li et al [11] have proposed a physical activity (PA) recognition algorithm for a wearable wireless sensor network using both ambulatory electrocardiogram (ECG) and accelerometer signals using support vector machine (SVM) and Gaussian mixture models (GMM). Ambulatory ECG signals/recordings have recently been used for several other purposes like classification of paroxysmal and persistent atrial fibrillation [12], automated recognition of obstructive sleep apnea syndrome [13], an embedded mobile ECG reasoning system for elderly patients [14], ECG signal compression and classification [15], heart rate and accelerometer data fusion for activity assessment [16], a patient adaptive profile scheme for ECG beat classification [17], automatic detection of respiratory rate [18], an intelligent telecardiology system to detect atrial fibrillation [19], etc.…”