Extracting evidence from patient narratives

To help turn the data in medical transcripts into insight for better patient care, Linguamatics and RealHealthData are partnering. The arrangement combines Linguamatics’ natural language processing (NLP) text analytics technology with RealHealthData’s extensive database of detailed provider narratives. The goal is to improve the understanding of drug use, adverse events and product switching in settings outside clinical trials.

According to the two companies, understanding the real-world impact of therapies on patients is essential for pharmaceutical and biotech firms. Medical records are a main source of real-world data and provide evidence important to all phases of drug development. RealHealthData offers access to patient narratives from every state in the United States and every medical specialty. Linguamatics 12E can be used to extract key facts from those narratives using relevant ontologies and queries.

Manuel Prado, CEO of RealHealthData, says, “Deploying Linguamatics I2E Advanced NLP engine to the RealHealthData database of detailed provider narratives is a natural fit. Current and future customers can now access the unique and valuable insights in the database using a first-in-class, healthcare-specific natural language processing platform.”

The unstructured text of electronic health records and patient records provides a level of detail and granularity not available from the structured fields with which life science companies usually work, according to the companies. RealHealthData’s database of patient narratives contains detailed records that include valuable information that could impact patient outcomes, such as patient social status and detailed clinical characteristics (comorbidities, complications, co-medications, lab values, adherence or switching issues).

David Milward, CTO of Linguamatics, says, “Using I2E, we can search this huge amount of unstructured patient data using NLP, incorporating machine learning to directly find patients of interest (e.g. diabetic patients who smoke and have a weight over 80kg) and use the longitudinal data to look at outcomes or behavior over time.”