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

Area of Science:

• Biomedical Engineering
• Dental Public Health
• Machine Learning Applications

Background:

• Oral health is integral to systemic health, linked to conditions like cardiovascular disease and diabetes.
• Proper toothbrushing is crucial for preventing dental caries and periodontal disease, yet adherence to correct techniques is often poor.
• This gap necessitates innovative solutions for monitoring and improving brushing habits.

Purpose of the Study:

• To develop and evaluate a fine-grained toothbrushing region recognition system.
• To assess the efficacy of machine learning classifiers and inertial measurement units (IMUs) for real-time brushing analysis.
• To enhance the reliability of oral hygiene monitoring through advanced signal processing.

Main Methods:

• Utilized six machine learning classifiers and two IMUs (toothbrush holder and wrist-mounted).
• Developed a hierarchical approach to identify brushing activities and recognize specific oral regions.
• Implemented post-processing strategies including contextual smoothing and majority voting for improved accuracy.

Main Results:

• The Random Forest classifier achieved the highest accuracy (96.13%), sensitivity (96.10%), precision (95.51%), and F1-score (95.60%).
• The proposed system effectively distinguished between brushing and transition activities.
• Demonstrated feasibility for detailed toothbrushing region recognition.

Conclusions:

• The developed approach provides effective and feasible fine-grained toothbrushing region recognition.
• This technology holds potential for improving toothbrushing monitoring and promoting better oral hygiene.
• Accurate recognition of brushing habits can contribute to the prevention of oral diseases.