The technology combines electronic medical record (EMR) data and data extracted with natural language processing (NLP) for use in protocol design and feasibility, as well as site selection and patient identification.
The company has seen "huge interest" from both healthcare organizations and biopharmaceutical companies for access to this type of data, David Fusari, CTO and co-founder of TriNetX told Outsouring-Pharma.com.
TriNetX also works with contract research organizations (CROs) such as Icon.
The company recently expanded its headquarters in Cambridge, MA in response to industry demand for its clinical trial services.
How it works
Clinical trial protocols frequently include criteria not captured as structured data elements in electronic health records (EHRs), Fusari explained. For example, information such as cancer stage and histology as well as treatment plans, pulmonology test results, and disease-specific performance status scores, may be included.
“TriNetX uses NLP to make these data elements available for query by extracting them from unstructured text, mapping them to standard clinical terms, and making them available to query through TriNetX Live,” Fusari explained.
The service is based on technology from Averbis, a Germany-based text-mining and machine-learning company, which has experience applying NLP in healthcare.
“It’s important that NLP understands negation, or language that indicates the absence of a condition. For example, if a clinical note states that the patient ‘denies having’ a disease or ‘has no history of a disease,’ NLP should not associate the disease with the patient,” said Fusari.
Additionally, NLP has to understand context, he said. For instance, information from a patient’s family history should not be associated specifically with the patient.
“TriNetX’s NLP service utilizes sophisticated algorithms and we actively monitor actual output to the expected output and conduct tuning when necessary,” Fusari added.