Wearable sensors

The long-term aim of this study is to investigate how age-related decreases in proprioceptive abilities and associated risk factors for falls and injury in the elderly can be compensated for by a biofeedback device. The biofeedback device would ultimately be used during daily functional activities to decrease morbidity and mortality associated with falls and dependency in the elderly.

The goal of the current pilot study is demonstrating that analysis of time-varying plantar pressure and acceleration patterns can create computational intelligence to identify situations where a person exhibits abnormal postural control.  This methodology for detecting abnormal postural control can then be used to design a biofeedback device to compensate for age-related proprioceptive loss.

This research project is conducted in collaboration with S. Zeigler and S.Marocco from Physical Therapy department of Clarkson University.

Our first prototype was manufactured in my lab using a pressure-sensitive insole that we designed and a wireless 2D accelerometer. Only one shoe was manufactured, through the plan is to use a pair ;-)

continue

If video above does not play, try this link

Gait Pattern Recognition

Using this prototype we were able to recognize 5 different activities performed by the subject:

1. Normal standing posture

2. Simulated geriatric standing with most of the body weight shifted to the heels

3. Normal gait

4. Simulated geriatric gait (less pronounced heel strike)

5. Simulated "tip-toe" walking

Normal standing

Geriatric standing

Normal gait

Simulated geriatric gait

Simulated forefoot gait

Currently we are working on collecting the data from human subjects using our second generation device. The motion of a heel in space is tracked using 6 degree of freedom wireless inertial tracker, comprised of a low-noise 3D accelerometer and a 3D gyroscope. The data are delivered wirelessly to a PC. The data samples are synchronized between multiple wireless devices on the order of a few microseconds. Plantar pressure distribution is captured using FScan mobile.

Using acquired data and techniques of computational intelligence we are planning to recognize and quantify various activities performed by subjects.

6DOF wireless sensor

Plantar pressure, heel and body sensors

PC interface to wireless data

Publications

·     Identification of gait types from plantar pressure and heel acceleration data, Sazonov ES, Bumpus T, Zeigler S, Marocco S, Proceedings of International Joint Conference on Neural Networks, Montreal, 2005.

12/2/2006