AI-Assisted Intelligent Toothbrush for Toothbrushing Habit Formation

Abstract

Oral diseases affect over 3.7 billion people globally, yet most are preventable with proper toothbrushing techniques. These techniques—characterized by brushing force, brush positioning, motion, coverage, and time—are highly personalized and difficult to adopt without timely feedback. This underscores the need for systems that can provide real-time brushing monitoring at personal level. A key challenge is assessing bristle–tooth contact quality, which is essential for ensuring comprehensive coverage and appropriate force application. This thesis presents an intelligent toothbrush designed for personalized technique monitoring. The toothbrush features a soft, pillar-like 3D force sensor and a contact-pattern classifier based on a Temporal Convolutional Network to evaluate bristle–tooth contact quality. The classifier achieves 98.58% accuracy across 10 brushing patterns, enabling monitoring of contact position, motion, and force. Leveraging the repetitive nature of brushing, a temporal-averaging method improves out-of-distribution detection, raising AUROC from 0.81 to 0.87 over a per-window baseline. In combination with an IMU-based posture classifier based on Deep Ensembles, a sensor-fusion framework enables fine-grain coverage recognition at two-teeth resolution. An interactive user interface further augments the system with gamified brushing guidance, delivering real-time feedback on key brushing parameters. Finally, the toothbrush is mounted on a robotic arm to demonstrate its ability to enable behavior adjustment. By integrating bristle-level sensing, AI-assisted recognition, and interactive feedback, the proposed system enable personalized toothbrushing monitoring for habit formation, with the potential to promote long-term oral health.

Date of Award	2025
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology
Supervisor	Yajing SHEN (Supervisor)