Principal Investigator: Professor Inke Konig
University of Lubeck (Germany)Tags: 48012, Machine Learning, Population Structure, risk prediction
One important aim of precision medicine is to estimate the individual risk for common diseases to enable effective prevention and early intervention. Current models for risk prediction are often created using data from one specific population. However, recent results show that even subtle differences in the underlying genetic architecture of different populations may distort the individual risk predictions. This is of major concern, if the respective person comes from a different population than the model was created on. This may lead to unnecessarily increased prevention efforts if the risk was estimated too high or to absence of prevention efforts if the risk was estimated too low, which in turn may lead to a lower quality of life. We are facing this situation especially in Europe, since migration from one European country to another was never easier than in these days.
In this six month project we first want to investigate how sensitive current risk prediction models are to different European populations. In a second stage we want to investigate how better risk prediction models can be constructed, taking the underlying population structure into account.