Comparing classification algorithms in predicting COVID19 deaths

In this work we compare a number of classification algorithms in predicting COVID19 deaths.  We combine data from a number of different datasets in the Vivli database and we demonstrate that while classic algorithms perform well in terms of accuracy, if one wands to increase the specificity it needs to treat the imbalance in the number of observations between the two classes. In almost all the cases improving specificity has a small cost in the overall accuracy.