Leveraging WiFi Sensing toward Automatic Recognition of Pain Behaviors

WiFi sensing has been well explored for recognizing human activity types. However, research is limited in the possibility of its use in identifying affective expressions such as behaviors associated with pain experience. As critical groundwork, we investigated the use of channel state information from WiFi devices for capturing speed and lateral asymmetry attributes of physical activity. These two attributes are body movement qualities associated with hesitation and guarding, respectively, which are pain behaviors that are valuable to address in physical rehabilitation for people with chronic pain. We obtained mean F1 scores of 0.92 and 0.90 for automatic detection of movement speed levels and lateral asymmetry. These findings suggest that WiFi sensors could be a valuable alternative or supplement to traditional motion capture systems, for unobtrusive, continuous evaluation for hesitation and guarding behaviors in everyday physical activity in the home.