By Topic

Regression models for estimating gait parameters using inertial sensors

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$33 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

5 Author(s)
Braveena K. Santhiranayagam ; School of Sport and Exercise Science, Victoria University Melbourne, Australia ; Daniel Lai ; Alistair Shilton ; Rezaul Begg
more authors

Advanced mathematical models are now widely used in medical applications for diagnosis, prognosis, and prevention of diseases. This work looks at the application of advanced regression models for estimating key foot parameters in falls prevention research. Falls is a serious issue for the rapidly increasing elderly demographic. We propose to investigate the notion of falls prediction through the use of portable, light weight, easy to use and inexpensive sensors along with advanced computational intelligence estimation models. This study compares two mathematical models namely the Generalized Regression Neural Networks (GRNN), and the Support Vector Machine (SVM) to estimate the key gait parameters. The study deployed Inertial Measurement Units (IMU) consisting of accelerometers and gyroscopes sensors to measure the foot kinematics and an optoelectronic motion capture system to validate the results. Our results demonstrated that both mathematical models estimate the key end point foot trajectory parameters (1) mx1 - first maximum after toe-off (root mean square error (rmse) range of 2.0 mm to 12.5 mm) (2) normalized time to mx1 (rmse range of 0.4% to 3.7%) and (3) Minimum Toe Clearance (rmse range of 2.0 mm to 10.2 mm) and (4) normalized time to MTC (rmse range of 0.7% to 5.4%) using IMU features. The SVM regressor showed better estimation rmse 56 times out of the 70 comparison estimations. In all cases the best model respectively from the GRNN and SVM family of models was compared.

Published in:

Intelligent Sensors, Sensor Networks and Information Processing (ISSNIP), 2011 Seventh International Conference on

Date of Conference:

6-9 Dec. 2011