Facial recognition to monitor the faces of ICU patients? It is the idea that came to a group of researchers at the hospital of the University of Yokohama, Japan. The research, presented at the congress of the European Society of Anaesthesiology, provides a fully automated and computerized system through which it is possible to monitor and perhaps even predict the unsafe behavior of patients in intensive care.
Unsafe behavior can be considered, for example, as the accidental removal of the respiratory tube. According to the first tests, the system has a fairly high precision (75% according to the researchers). The monitoring is done by means of a camera mounted on the ceiling above the bed where the patient is hospitalized.
A special automatic learning algorithm was instructed by the researchers through hundreds of hours of recording collected during the first phase, an algorithm that then proved useful in recognizing high-risk behavior in the first tests. It is a system that could make up for cases of limited personnel or that could prove useful in all those cases in which the staff itself can not continuously, 24 hours a day, monitor unconscious patients.
“We were surprised by the high degree of accuracy achieved, which shows that this new technology has the potential to be a useful tool to improve patient safety and is the first step towards an intelligent ICU (intensive care unit) planned in our hospital,” says Akane Sato, one of the researchers involved in the project.