Using Machine Learning to Predict Nosocomial Infections and Medical Accidents in a NICU

The final sample thus includes 5,466 neonates and represents over 101,621 infant/days over the sample period.3 Table 1 provides the means of the main variables used in the models. [...] The table shows that proportionately more MA occurred in the months that followed the reform relative to either the popula- tion of neonates or to those with a HCAI. [...] The analysis of MA fur- ther includes daily regular and overtime hours in the NICU as well as a dummy variable equal to one if the infant incurred a HCAI prior to the MA. [...] The first number inside each node corresponds to the predicted probability of reaching the latter.5 The number of observations (n) in appears just below the node. [...] The main idea is to fit a tree after removing the random effects part of the model, update the estimates (or predictions) of the random effect and cycle until convergence.