iBrush: Toothbrushing Monitoring using Smartwatch

Daily toothbrushing is an essential habit for preventing dental diseases. However, existing technologies to monitor the effectiveness of toothbrushing are very limited. In this paper, we present the design of iBrush, a system that can detect incorrect brushing techniques, localize brushing surfaces, and estimate brushing duration on each surface using an off-the-shelf smartwatch during manual toothbrushing. iBrush employs a novel toothbrush design in which small magnets are attached to the handle of a toothbrush so that its orientation and motion can be captured by the magnetic sensor in the user's smartwatch. iBrush also uses the inertial sensor on the smartwatch to classify different types of arm/wrist brushing movements. Combining the toothbrush orientation and the hand movement patterns, our system can recognize the effectiveness of toothbrushing. Acoustic signals collected from the smartwatch and user-specific toothbrushing order are also utilized to improve recognition accuracy. We implemented the proposed system on a commercial smartwatch. In extensive experiments with 12 users over three weeks, our system successfully incorrect toothbrushing brushing techniques with an F1 score of 90.5% and localized the sixteen brushing surfaces with a F1 score of 85.6%.