Ranking Clinical Trials Using Elasticsearch

The clinical trials task of the TREC 2017 Precision Medicine Track was designed to represent the potential for connecting patients with experimental treatments if existing treatments were ineffective. Participants were challenged with the task of retrieving appropriate clinical trials from ClinicalTrials.gov for which a patient is eligible. This paper presents an approach to solving the problem by first preparing an index for the clinical trial descriptions based on specific tags in the XML files and querying them using Elasticsearch. Initial results indicate that our approach performed very well for certain kinds of queries – however, more tuning may be required for ensuring generalizable results from the search.