Link Prediction Approaches for Disease Networks

In the last years link prediction in complex networks has attracted an ever increasing attention from the scientific community. In this paper we apply link prediction models to a very challenging scenario: predicting the onset of future diseases on the base of the current health status of patients. To this purpose, a comorbidity network where nodes are the diseases and edges represent the contemporarily presence of two illnesses in a patient, is built. Similarity metrics that measure the prox- imity of two nodes by considering only the network topology are applied, and a ranked list of scores is computed. The higher the link score, the more likely the relationship between the two diseases will emerge. Exper- imental results show that the proposed technique can reveal morbidities a patient could develop in the future.