Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Multimodal physical activity recognition by fusing temporal and cepstral information.

Ming Li1, Viktor Rozgica, Gautam Thatte

  • 1Viterbi School of Engineering, University ofSouthern California, Los Angeles, CA 90089, USA. mingli@usc.edu

IEEE Transactions on Neural Systems and Rehabilitation Engineering : a Publication of the IEEE Engineering in Medicine and Biology Society
|August 12, 2010
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

A dynamic Bayesian network approach to modeling engagement and walking behavior: insights from a yearlong micro-randomized trial (<i>Heartsteps II</i>).

Health psychology and behavioral medicine·2025
Same author

Individual, Social and Environmental Influences on Sleep in Latino Pre-Adolescents: A Qualitative Analysis.

Journal of adolescent research·2025
Same author

Dynamic Modeling and System Identification of User Engagement in mHealth Interventions using a Bayesian Approach for Missing Data Imputation.

Control engineering practice·2025
Same author

Beyond the current state of just-in-time adaptive interventions in mental health: a qualitative systematic review.

Frontiers in digital health·2025
Same author

Modeling engagement with a digital behavior change intervention (HeartSteps II): An exploratory system identification approach.

Journal of biomedical informatics·2024
Same author

The frequency of using wearable activity trackers is associated with minutes of moderate to vigorous physical activity among cancer survivors: Analysis of HINTS data.

Cancer epidemiology·2023

This study introduces a novel physical activity (PA) recognition algorithm using electrocardiogram (ECG) and accelerometer data. The advanced system significantly improves accuracy compared to existing methods.

Area of Science:

  • Biomedical Engineering
  • Signal Processing
  • Wearable Technology

Background:

  • Accurate physical activity (PA) recognition is crucial for health monitoring.
  • Wearable wireless sensor networks offer a promising platform for continuous PA tracking.
  • Integrating electrocardiogram (ECG) and accelerometer data can enhance recognition accuracy.

Purpose of the Study:

  • To develop and evaluate a multimodal and multidomain algorithm for physical activity recognition.
  • To leverage both time-domain and cepstral features from ECG and accelerometer signals.
  • To fuse information from different domains and sensors for improved performance.

Main Methods:

  • Utilized Hermite polynomial expansion and principal component analysis for ECG signal analysis.

Related Experiment Videos

  • Extracted time-domain features from ECG and accelerometer data, classified using Support Vector Machines (SVM).
  • Modeled cepstral features using Gaussian Mixture Models (GMMs) and reduced dimensionality with heteroscedastic linear discriminant analysis, fusing multimodal and multidomain subsystems at the score level.

    • Main Results:

    • Achieved classification accuracy ranging from 79.3% to 97.3% across various scenarios.
    • Demonstrated a relative error reduction of over 24% compared to state-of-the-art single accelerometer systems.
    • Validated the effectiveness of multimodal and multidomain fusion for PA recognition.

    Conclusions:

    • The proposed algorithm effectively recognizes physical activity by integrating ECG and accelerometer data.
    • Multimodal and multidomain fusion significantly enhances the performance of wearable-based PA recognition systems.
    • This approach offers a robust and accurate solution for health monitoring and activity tracking.