Classification of Handwashing and Similar Activities.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Wearable Technology

Background:

• Handwashing is crucial for preventing germ transmission.
• Distinguishing handwashing from similar activities is challenging.
• Wearable sensors offer a novel approach to monitoring hygiene behaviors.

Purpose of the Study:

• To develop and evaluate a machine learning model for classifying handwashing from similar activities using wearable sensor data.
• To assess the accuracy of an ensemble gentle boost decision tree algorithm for this classification task.
• To explore the clinical relevance of wearable sensors in promoting proper hand hygiene.

Main Methods:

• Data collected using a 3-axis accelerometer and gyroscope worn on the wrist.
• Feature identification and creation of a feature matrix.
• Classification using MATLAB's classification learner app with an ensemble boosted classifier.

Main Results:

• An optimized boosted ensemble classifier achieved an overall classification rate of 91.6%.
• The algorithm successfully differentiated handwashing from activities like applying lotion.
• Demonstrated the feasibility of using wearable sensors for hygiene monitoring.

Conclusions:

• Wearable sensors combined with machine learning can accurately detect handwashing.
• This technology has the potential to alert users and guide proper hand hygiene practices.
• Contributes to infection prevention strategies, especially during pandemics.