Facial Recognition Continuously Monitors ICU Patients

After collecting about 300 hours of daytime image data of patients facing the camera in body positions that showed their face and eyes clearly, 99 images were subject to a machine-learning (ML) algorithm to analyze them. Based on input from the observational data, especially from around the subject's face, the ML algorithm learned to identify potential high-risk behavior in a process that resembles the way a human brain learns new information.

In a proof of concept study that included 24 postoperative patients (average 67 years of age) who were admitted to the ICU in Yokohama City University Hospital between June and October 2018, the ML algorithm was able to identify high risk unsafe behavior--such as accidentally removing their breathing tube--with 75% accuracy. They also suggested that monitoring consciousness may improve accuracy by helping to distinguish between high-risk behavior and voluntary movement. The study was presented at the Euroanaesthesia annual congress, held during June 2019 in Vienna (Austria).

“Using images we had taken of a patient's face and eyes we were able to train computer systems to recognize high-risk arm movement,” said lead author and study presenter Akane Sato, MD. “We were surprised about the high degree of accuracy that we achieved, which shows that this new technology has the potential to be a useful tool for improving patient safety, and is the first step for a smart ICU which is planned in our hospital.”

Facial recognition systems use biometrics to map facial features from a photograph or a video. The geometry of the face is then analyzed, with key factors including interpapillary distance and the distance from forehead to chin. In all, there are over 65 quantifiable features that can be used to identify a face, generating a unique facial signature.

Related Links:
Yokohama City University