Smart Teeth: Sensor-Embedded Teeth for Oral Activity Recognition

Cheng-Yuan Li1
Yen-Chang Chen1
Wei-Ju Chen1
Polly Huang2
Hao-hua Chu1
Department of Computer Science and Information Engineering1,
Department of Electrical Engineering2
National Taiwan University, Taipei, Taiwan
{d99922035, b98902090, r99922148}@csie.ntu.edu.tw, phuang@cc.ee.ntu.edu.tw, hchu@csie.ntu.edu.tw



We designed and developed an oral sensory
system that can recognize human oral activities. Our results
from a laboratory experiment with 8 participants
demonstrate the feasibility of this oral sensory system in
recognizing the following four human oral activities:
speaking, chewing, drinking, and coughing. We found that
a person-dependent SVM classifier achieved a high F-
measure accuracy of 93.8%, whereas a person-independent
SVM classifier achieved only an F-measure accuracy of
59.8%.
Because the mouth is an opening into human health, this
oral sensory system has the potential to enhance exiting
oral-related healthcare monitoring applications such as
dietary tracking. We will explore several future work
directions. First, we will design the next prototype, which
will be integrated with wireless communication and battery-
recharging capabilities. Second, we would like to improve
the accuracy of our system’s activity classification. Finally,
we will continue working with dental collaborators to
improve the safety of our system.