The Patient-Centered Outcomes Research Institute granted $1 million to Boston-based Brigham and Women's Hospital to build natural language processing technology that can extract information from EHRs, according to a Nov. 20 news release.
Through the project, Brigham and Women's Hospital researchers aim to develop NLP-based technology that can address the challenge of extracting information when the topic or command is switched mid-sentence.
Researchers will focus on two different NLP technologies: artificial intelligence technology based on deep learning, and Canary software, which uses programming code changes. The team will build these technologies based on three clinical examples: patients with high cholesterol and high risk for heart attacks or stroke who decline statin cholesterol-lowering medications, patients with obesity who discuss weight-loss surgery with their clinicians and recording of the patient's smoking status in the EHR.
Brigham researchers will then evaluate the technologies' performances using EHR data from Baltimore-based Johns Hopkins Health System to ensure the technology works well on data from different organizations.