Publication Type:Journal Article
Source:Biomed Data J., Volume 1, Issue 2, p.10-12 (2015)
Keywords:Drug Repositioning Benchmark Dataset Text Mining Computational Methods
As the amount of the publicly available scientific literature increases, efficient text mining techniques are required to aid the biomedical knowledge workers in extracting the most important information from text and, ideally, to train models that may automatically suggest novel hypotheses. In this latter direction, the perspective of the employment of text mining techniques for computational drug repositioning, i.e., predicting new indications for existing drugs, is constantly attracting attention. One of the main obstacles for the development and establishment of such techniques is the systems’ evaluation, as there is lack of benchmark datasets for performance evaluation. In this paper we introduce such a dataset for the evaluation of systems that perform computational drug repositioning. The dataset comprises 54 drug repositioning cases, the information for which was manually compiled, curated and integrated.
The new indications for the reported cases have been approved by FDA or EU RUS in the period 1955-2013.