NLP engine aims to improve patient outcomes through efficient data review

NLP-engine-to-improve-patient-outcomes-through-efficient-trial-review.jpg
(Image: Getty/metamorworks) (Getty Images/iStockphoto)

LabKey and Linguamatics have designed an integrated NLP data management solution to help accelerate clinical data abstraction and curation of unstructured notes and reports for clinical research.

The natural language processing (NLP) text analytics provider Linguamatics and LabKey, a provider of bioinformatics data management solutions, aim to reduce the manual processes involved in clinical data and information analysis with Lingumatics’ NLP engine I2E and LabKey Server’s document processing and curation user interface.

Simon Beaulah, senior director of health care at Linguamatics, told us that the two companies began working together as part of a project at the National Cancer Institute, which established an NLP pipeline that automates pathology report annotation and review to build a “gold standard” for machine learning training.

The work involved using Linguamatics I2E to identify information in thousands of pathology reports from different states, which was followed by a curation process in LabKey,” said Beaulah.

The companies believe that per the collaboration the use of I2E and the use of LabKey can extract important drug discovery and clinical concepts presented for review as to make chart reading more efficient for clinical development.

The integrated platform enables companies to extract electronic health record (EHR) data into cancer registries and clinical data warehouses, reducing the manual burden associated with clinical study review. Subsequently, relevant population health data and metrics can be used to improve quality metrics reporting and patient outcomes, according to the company.

Beaulah further explained, “The use of I2E to extract insights from text and then have the results curated in LabKey supports an augmented intelligence approach that ensures manual review is focused on the specific information of interest and not the whole document.”

Companies like TriNetX have said that there has been a “huge interest” for access to data extraction through NLP’s for use in protocol design and even site selection and patient identification.