Discovering Commonly Shared Semantic Concepts of Eligibility Criteria for Learning Clinical Trial Design

摘要

Discovering commonly shared semantic concepts of eligibility criteria can facilitate hospitals in recruiting appropriate target population and empower patients with more effective result ranking of concept-based search, as well as assist researchers in understanding clinical trial design. This study aims to identify commonly shared semantic concepts of eligibility criteria through the identification of eligibility criteria concepts for each disease. An automated approach for extracting semantic concepts from eligibility criteria texts is proposed. For each disease, commonly shared semantic concepts are determined for reviewing the commonly shared concepts of clinical trials. Our experiment dataset are 145,745 clinical trials associated with 5,488 different types of diseases on ClinicalTrials.gov. 5,508,491 semantic concepts are extracted with 459,936 being unique. We further analyze its application on assisting researchers in learning disease-specific clinical trial design.