The Danish startup Teton has used Gefion, one of the most powerful supercomputers in Europe, to draw the development of his AI “Care Companion” for hospital staff.
Teton installs cameras and sensors in hospital rooms to collect real -time data. This is transferred to an AI algorithm and creates a virtual “digital twin” of the room. The model monitors the behavior of the patient and personnel such as exercise, breathing or changes in posture. When it sees a problem, the system alerts nurses via an app.
In order to protect the privacy, the entire processing takes place in front of the device and nobody is sent to the cloud. No personal data or raw video film material is saved. Instead, the video is immediately converted into 3D spatial representations and the original film material is deleted.
Teton's goal is to help employees in the hospital to recognize signs of health problems at an early stage and to react faster to potential risks. The company also aims to facilitate the workload of the nurses. By continuous monitoring patient rooms, the system can reduce the number of manual checks. It can also automatically generate maintenance.
European Tech superstar
Connected by founders, investors and innovation champions at the TNW conference on the 1000s on June 19 and 20.
The AI care manager has already been tested in two Danish hospitals with promising results. According to the company's own research, the tool has reduced the night load of Nightshift -Workloads by up to 25%.
Teton's founder, Mikkel Wad Thorsen and Esben Klint Thorius, I spent years to train your AI model for real patient scenarios. They even hired actors and nurses to repeatedly affect scenes to improve their training data. However, these manual processes take a lot of time. Fioned there, Denmark Ai -Supercomputercame in.
Use the Teton supercomputer was able to create over 1 million 3D scenes from hospital rooms. Another scenario would take place in every virtual space. For example, a patient turns in bed, a nurse who checks vital, or someone with need.
These scenes are not only static images, but also dynamic environments that imitate how people and objects move and behave in real life. By producing so many of them, Teton can expose his AI to a variety of scenarios – much more than manually realizable. This helps the system learn how to recognize subtle patterns and react appropriately in real clinical environments.
“Gefion made a big difference in the rate with which we were able to develop our AI model.” Thorsen customs tnw.
“The possibility of unlocking geometric knowledge helps us to predict much better predictions and estimates about how one patient is doing,” he added.
Thorsen In the end, hopes that improvements will help to ensure that hospitals ensure better patient care – and take care of the weight of the nursing staff.
Give helping hands
The EU currently has 1.2 million doctors, nurses and midwives. According to the OECD, there is also a declining interest in care in nursing in more than half of the 27 countries of the Bloc Research.
The shortage of care is a worldwide topic that has led to several startups that try to close the gap with technical solutions. One of them is Bemlo in Stockholm, which has created a platform for the short -term hiring of nurses and doctors and enables hospitals to quickly bring new employees on board. Other efforts try to support nurses more directly. The hard -working robotics based in the USA built Moxi, one meter Mobile robot that should help to alleviate boring and repeating tasks in health facilities.
Armed with supercomputing -armed and digital twins, Teton Wetten can bet that AI makes it easier to burden the extended hospital workers. And now the startup wants to quickly follow the development of its algorithm. “With all these new data points, we were able to speed up the iteration time considerably and unlock a new scale for the next generation of our model” Thorsen.
