Categories and clusters to investigate similarities in Diabetic Kidney Disease patients

Heterogeneous responses to therapeutical treatments across patients and over time is a common and serious problem for several diseases. Precision medicine research focuses in developing procedures to take treatment decisions for the individual patient using all the information available for the patient, including demographic and clinical variables and the response to the followed treatment. In this paper we adopt category theory and the cluster analysis to achieve insight into specific disease pathways and patient subgroups. We analyze a longitudinal dataset of patients affected by diabetic kidney disease (highly prevalent in type 2 diabetes) and monitored at different time points in the response to various treatment regimes. This analysis, based on distances between patients in different time points and in time evolution, divides patients into clusters that show the relevant role of some variables in affecting the progress of the disease.