Recommending Personalized Treatment for Patients

This is an under developement website, where medical doctors can have the opportunity to get drug recommendations for their patients and explore the similarity among different clinical cases of their patients. Our recommendation system runs on multi-dimensional graph data consisting of different participating entities (i.e. patients, drugs, diseases, etc.). In particular, our system runs on a Heterogeneous Information Network (HIN) and uses meta paths to infer similarities among entities. A meta path is a sequence of different node and edge types, that carry a different semantic message about the relation among graph entities (Patients, Drugs, and Diseases).

You can read more details in our paper about our COVID-19 research by clicking here.

The network schema of our Health Graph is shown in the following figure.