Risk of pneumonia and associated outcomes in intensive care unit: An integrated approach of visual and cluster analysis

Big data analysis may contribute to widen information for the prevention of diseases by the identification of risk factors and the prediction of outcomes, such as healthcare associated infections (HAIs) in Intensive Care Units (ICUs). Here, for the first time, we present an integrated approach of visual and cluster analysis, supported by traditional statistical methods, to identify and to describe determinants of risks of pneumonia and associated adverse outcomes in ICU patients. Our study indicates that Sankey diagrams are useful tools to visualise flows of patients from their admission to ICU and how each cluster and duration of intubation contribute to the acquisition of pneumonia. Moreover, they enabled us to graphically represent the role of Acinetobacter baumannii, Klebsiella pneumoniae and Pseudomonas aeruginosa-associated pneumonia on the risk of sepsis and death. The use of an integrated approach of cluster, visual and statistical analyses allows a better understanding, interpretation and communication of health data on specific issues in the field of Public Health.