Navigation auf uzh.ch
Navigation auf uzh.ch
DSI Infrastructures & Labs are shareable infrastructures or structural vessels for creating collaborative research environments related to digital transformation. Prof. Dr. Christof Seiler briefly introduces RAI, one of the projects in this series.
What is RAI?
RAI stands for reliable AI. It is a software platform that makes it possible to calculate how reliable modern AI algorithms are. We worked with a professional cycling team. We built an algorithm that can predict how many calories an athlete consumes during a race. Based on our calculations, the coaches were able to put together optimal meals.
Why is such a tool needed?
Our calorie calculation algorithm is not perfect and makes mistakes, as do other AI algorithms. That's why we have calculated a whole range of possibilities. For example, instead of predicting that a Tour de France athlete will consume 4500 calories in the second stage, we calculate that an athlete will consume between 4400 and 4600 calories. We believe that such prediction intervals are very important in biomedicine.
How will biomedical researchers be able to use this tool?
Many biomedical researchers develop algorithms that make predictions. For example: Which drug should we try to treat a disease? Such predictions can be estimated based on past experiments. It may be that you simply see an event by chance. RAI enables researchers to make their prediction models more reliable and to provide them with prediction intervals.
All projects of the series «DSI Infrastructures & Labs» can be found here.
Prof. Dr. Christof Seiler is a group leader at the Department of Rheumatology at USZ and UZH and an assistant professor of statistics at the Department of Advanced Computing Sciences at Maastricht University. His group's focus is on biomedical data science.
