How are AI based Personalized Risk Models trained?

We will start with a collection of longitudinal data of infected patients from clinical routine as well as clinical studies.
Based on these data we will train several machine learning models. These models will allow us to forecast for an individual patient his / her risk for a number of future events, such as hospitalization, admission to Intensive Care Units, mechanical ventilation, acute respiratory failure, sepsis or mortality.
With the help of recent methods, we will make individual models explainable and thus enable a transparent decision support tool.