Wristalert: A Face Touching Alert Through Inertial Sensors and Long Short-Term Memory Neural Network

A wearable face-touching prevention device intended for diminishing the contagion risk from viruses and other pathogens has been developed, named wristalert. The purpose is to identify and emit an alert when face-touching behaviors are detected. The wristalert draws awareness to subconscious hand movements toward the face, particularly when a person is exposed to high-risk places where a contaminated surface could be reached, and self-inoculation may occur. It is based upon a trained neuronal network using inertial sensors and implemented into a microcontroller to drive a sound alert when wrist-hand movements may reach the face. The electronic components are embedded in a 3-D-printed case attached to a strap, forming the alerting device. The results show that the wristalert is highly reliable since an 89% prediction accuracy is obtained, providing a feasible system to avoid face touching. This wearable device is replicable, inexpensive, and has low power consumption.