Machine Learning-Based Toothbrushing Region Recognition Using Smart Toothbrush Holder and Wearable Sensors

Oral health is a critical factor in maintaining overall health, and its association with systemic diseases, including cardiovascular disease and diabetes mellitus, has been extensively investigated. Effective plaque removal through proper toothbrushing techniques is fundamental for preventing dental caries and periodontal diseases. Despite standardized guidelines, many individuals fail to adhere to correct brushing techniques, thereby increasing the risk of oral diseases. To address this issue, this study proposes a fine-grained toothbrushing region recognition approach incorporating six machine learning classifiers and two inertial measurement units (IMUs), which are embedded in the toothbrush holder and mounted on the right wrist of the participant, respectively. By analyzing the continuous motion signals, the proposed hierarchical approach is capable of identifying brushing and transition activities and subsequently recognizing specific toothbrushing regions based on the predicted brushing activities. To further improve recognition reliability, post-processing strategies such as contextual smoothing and majority voting are applied. Experimental results demonstrate that random forest achieves the highest recognition accuracy of 96.13%, sensitivity of 96.10%, precision of 95.51%, and F1-score of 95.60%. The results indicate that the proposed approach is both effective and feasible for providing fine-grained toothbrushing region recognition in toothbrushing monitoring.