skip to main content

Title: Activity Segmentation Using Wearable Sensors for DVT/PE Risk Detection
Using a wearable electromyography (EMG) and an accelerometer sensor, classification of subject activity state (i.e., walking, sitting, standing, or ankle circles) enables detection of prolonged "negative" activity states in which the calf muscles do not facilitate blood flow return via the deep veins of the leg. By employing machine learning classification on a multi-sensor wearable device, we are able to classify human subject state between "positive" and "negative" activities, and among each activity state, with greater than 95% accuracy. Some negative activity states cannot be accurately discriminated due to their similar presentation from an accelerometer (i.e., standing vs. sitting); however, it is desirable to separate these states to better inform the risk of developing a Deep Vein Thrombosis (DVT). Augmentation with a wearable EMG sensor improves separability of these activities by 30%.
; ; ; ;
Award ID(s):
Publication Date:
Journal Name:
2019 IEEE 43rd Annual Computer Software and Applications Conference (COMPSAC)
Page Range or eLocation-ID:
477 to 483
Sponsoring Org:
National Science Foundation
More Like this
  1. Parkinson’s Disease (PD) is a neurodegenerative disorder affecting the substantia nigra, which leads to more than half of PD patients are considered to be at high risk of falling. Recently, Inertial Measurement Unit (IMU) sensors have shown great promise in the classification of activities of daily living (ADL) such as walking, standing, sitting, and laying down, considered to be normal movement in daily life. Measuring physical activity level from longitudinal ADL monitoring among PD patients could provide insights into their fall mechanisms. In this study, six PD patients (mean age=74.3±6.5 years) and six young healthy subjects (mean age=19.7±2.7 years) weremore »recruited. All the subjects were asked to wear the single accelerometer, DynaPort MM+ (Motion Monitor+, McRoberts BV, The Hague, Netherlands), with a sampling frequency of 100 Hz located at the L5-S1 spinal area for 3 days. Subjects maintained a log of activities they performed and only removed the sensor while showering or performing other aquatic activities. The resultant acceleration was filtered using high and low pass Butterworth filters to determine dynamic and stationary activities. As a result, it was found that healthy young subjects performed significantly more dynamic activities (13.2%) when compared to PD subjects (7%), in contrast, PD subjects (92.9%) had significantly more stationary activities than young healthy subjects (86.8%).« less
  2. 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 resultsmore »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.« less
  3. Abstract: In the past few years, smart mobile devices have become ubiquitous. Most of these devices have embedded sensors such as GPS, accelerometer, gyroscope, etc. There is a growing trend to use these sensors for user identification and activity recognition. Most prior work, however, contains results on a small number of classifiers, data, or activities. We present a comprehensive evaluation often representative classifiers used in identification on two publicly available data sets (thus our work is reproducible). Our results include data obtained from dynamic activities, such as walking and running; static postures such as sitting and standing; and an aggregatemore »of activities that combine dynamic, static, and postural transitions, such as sit-to-stand or stand-to-sit. Our identification results on aggregate data include both labeled and unlabeled activities. Our results show that the k-Nearest Neighbors algorithm consistently outperforms other classifiers. We also show that by extracting appropriate features and using appropriate classifiers, static and aggregate activities can be used for user identification. We posit that this work will serve as a resource and a benchmark for the selection and evaluation of classification algorithms for activity based identification on smartphones.« less
  4. Recent advances in machine learning and deep neural networks have led to the realization of many important applications in the area of personalized medicine. Whether it is detecting activities of daily living or analyzing images for cancerous cells, machine learning algorithms have become the dominant choice for such emerging applications. In particular, the state-of-the-art algorithms used for human activity recognition (HAR) using wearable inertial sensors utilize machine learning algorithms to detect health events and to make predictions from sensor data. Currently, however, there remains a gap in research on whether or not and how activity recognition algorithms may become themore »subject of adversarial attacks. In this paper, we take the first strides on (1) investigating methods of generating adversarial example in the context of HAR systems; (2) studying the vulnerability of activity recognition models to adversarial examples in feature and signal domain; and (3) investigating the effects of adversarial training on HAR systems. We introduce Adar, a novel computational framework for optimization-driven creation of adversarial examples in sensor-based activity recognition systems. Through extensive analysis based on real sensor data collected with human subjects, we found that simple evasion attacks are able to decrease the accuracy of a deep neural network from 95.1% to 3.4% and from 93.1% to 16.8% in the case of a convolutional neural network. With adversarial training, the robustness of the deep neural network increased on the adversarial examples by 49.1% in the worst case while the accuracy on clean samples decreased by 13.2%.« less
  5. Le, Khanh N.Q. (Ed.)
    In current clinical settings, typically pain is measured by a patient’s self-reported information. This subjective pain assessment results in suboptimal treatment plans, over-prescription of opioids, and drug-seeking behavior among patients. In the present study, we explored automatic objective pain intensity estimation machine learning models using inputs from physiological sensors. This study uses BioVid Heat Pain Dataset. We extracted features from Electrodermal Activity (EDA), Electrocardiogram (ECG), Electromyogram (EMG) signals collected from study participants subjected to heat pain. We built different machine learning models, including Linear Regression, Support Vector Regression (SVR), Neural Networks and Extreme Gradient Boosting for continuous value pain intensitymore »estimation. Then we identified the physiological sensor, feature set and machine learning model that give the best predictive performance. We found that EDA is the most information-rich sensor for continuous pain intensity prediction. A set of only 3 features from EDA signals using SVR model gave an average performance of 0.93 mean absolute error (MAE) and 1.16 root means square error (RMSE) for the subject-independent model and of 0.92 MAE and 1.13 RMSE for subject-dependent. The MAE achieved with signal-feature-model combination is less than 1 unit on 0 to 4 continues pain scale, which is smaller than the MAE achieved by the methods reported in the literature. These results demonstrate that it is possible to estimate pain intensity of a patient using a computationally inexpensive machine learning model with 3 statistical features from EDA signal which can be collected from a wrist biosensor. This method paves a way to developing a wearable pain measurement device.« less