Computer Vision Based Models for Psychomotor Skills
Maintaining skill at delicate medical procedures requires continued practice. This becomes challenging for procedures that occur infrequently but are required in emergent situations. Through a partnership with Carnegie Mellon University, researchers at the University of Pittsburgh are exploring new technologies to help teach providers these procedures and help them maintain these skills.
One possibility is through the use of various computer vision techniques that can analyze videos of a procedure being performed. These videos provide clues to the skill level of the person performing the procedure and may indicate the need for additional training. Below is an example of a video acquired using a video laryngoscope during a procedure known as endotracheal intubation (placing a breathing tube in the trachea).
A computer can watch these videos and analyze how different parts of the anatomy move with respect to the camera. This computer program then learns which movement patterns are likely to be successful and which are not to help guide providers through the procedure. In the future, new mannequins and simulators may use these technologies to help improve how we teach and continue to train critical skills to health care provider.