One of the most important challenges in the scenario of COVID-19 is to design and develop decision support systems that can help medical staff to identify a cohort of patients that is more likely to have worse clinical evolution. To achieve this objective it is necessary to work on collected data, pre-process them in order to obtain a consistent dataset and then extract the most relevant features with advanced statistical methods like principal component analysis. As preliminary results of this research, very influential features that emerged are the presence of cardiac and liver illnesses and the levels of some inflammatory parameters at the moment of diagnosis.
Keywords: COVID-19; feature extraction; imputation of data; principal component analysis; pseudo-anonymous data.