Parsable Clinical Trial Eligibility Criteria Representation Using Natural Language Processing.

Journal: AMIA ... Annual Symposium proceedings. AMIA Symposium
PMID:

Abstract

Successful clinical trials offer better treatments to current or future patients and advance scientific research. Clinical trials define the target population using specific eligibility criteria to ensure an optimal enrollment sample. Clinical trial eligibility criteria are often described in unstructured free-text which makes automation of the recruitment process challenging. This contributes to the long-standing problem of insufficient enrollment of clinical trials. This study uses a machine learning approach to extract clinical trial eligibility criteria, and convert them into structured queryable formats using descriptive statistics based on medical entity frequency and binary entity relationships. We present a JSON-based structural representation of clinical trials eligibility criteria for clinical trials to follow.

Authors

  • Jeongeun Kim
    College of Nursing, Seoul National University, Seoul, Republic of Korea.
  • Mitchell Izower
    Harvard Medical School, Boston, MA.
  • Yuri Quintana
    Harvard Medical School, Boston, MA.