Mining Online Heterogeneous Healthcare Networks for Drug Repositioning

Abstract

Drug repositioning represents the application of known drugs for new indications and plays an important role in healthcare research and industry. With its increasing value in drug development, multiple approaches have been applied in its exercise, basically classified as drug-based and diseasebased approaches. Our study adopted a disease-based approach and utilized Adverse Drug Reactions (ADRs) as an intermediary to associate drugs with diseases for drug repositioning. Based on the collected health-related social media data, we constructed a heterogeneous healthcare network and developed three path-mining techniques to identify significant associations between ADRs and diseases and then determine the associations between potential drugs and diseases for repositioning. When an ADR has a strong association with a disease based on the drugs indicated for the disease, there is an underlying mechanism-of-action (MOA) between the disease and the ADR. The ADR can be considered as a clinical phenotypic biomarker of the disease. Others drugs that have a strong association with the ADR are prospective for repositioning [12]. The experiment results demonstrate the repositioning capability of the proposed method and the advantages of using social media data. The case studies and the ratio of supporting articles show its effectiveness and the potential for further drug development research.