Identifying Patients for Clinical Trials Using Combined Data and Natural Language Processing

Global health research company TriNetX has announced they’ll be providing a natural language processing (NLP) service to healthcare organizations, biopharmaceutical companies and contract research organizations (CROs). The service combines electronic medical record (EMR) data with information mined from physician notes and clinical reports to help identify prospective patients for enrollment into clinical trials.

“With the TriNetX NLP service, we are able access data that our researchers have been very interested in for some time,” said Dr. Jack London, Informatics Core Director, Sidney Kimmel Cancer Center at Jefferson and Professor of Cancer Biology, Thomas Jefferson University. “Extracting this data from our clinical text reports helps our investigators better define and identify patient cohorts, and provides a larger data set for our academic and industry clinical research collaborations.”

The TriNetX platform allows users to analyze de-identified patient data in real-time, which could also help inform decisions regarding protocol design, site selection and study feasibility. Users can contact traceable healthcare organizations in order to build cohorts of patients that can be re-identified and recruited in clinical trials.

TriNetX combines structured and unstructured data to help clinical investigators identify candidates for clinical trials. Diagnoses, medications, lab results and genomics data are all examples of structured data, while unstructured data sources such as Corrected QT Interval for cardiac studies, helps to round-out the participant identification system.

“The combination of structured EMR data and NLP-extracted data is more powerful than either data set alone,” said Alex Eastman, Senior Director of Product Management at TriNetX. “NLP helps fill gaps in structured EMR data, and vice versa. You end up with a richer data set that supports advanced analyses.”