A team of scientists from Washington University, Carnegie Mellon University, Makerere University (Uganda) and Toyota Research Institute have developed a system that can detect and prevent medication errors in hospital patients. In the study, 13 anesthesiologists wore GoPro cameras for 55 days, capturing 418 medication pickups in various settings. The system used the captured videos to train a deep learning algorithm that recognized the visual characteristics of vials and syringe labels. As a result, the camera was able to identify in real time what the physician was holding. It was able to identify the contents of the vial and then check it against the label. In the tests performed, the system was 99.6% accurate in confirming matches and 98.8% accurate in detecting errors. According to the authors of the project, this technology could be an important step towards improving patient safety in medical facilities.