Improving Methods for Identifying Social, Behavioral, and Clinical Factors in Doctors' Notes in Electronic Health Records

This research project is in progress. PCORI will post the research findings on this page within 90 days after the results are final.

One of PCORI’s goals is to improve the methods that researchers use for patient-centered outcomes research. PCORI funds methods projects like this one to better understand and advance the use of research methods that improve the strength and quality of comparative effectiveness research.

What is the project about?

Many factors in patients’ lives can affect their health. For example, social factors like education or income and behavioral factors like smoking affect patients’ cancer risk. Electronic health records, or EHRs, include useful information about these factors, often in doctors’ notes. But researchers can’t easily study information from these notes because notes aren’t written in a standard way.

In this study, the research team is creating methods to identify social, behavioral, and clinical factors in doctors’ notes for use in research. The new methods use natural language processing, or NLP. In NLP, computer programs interpret written language and make it easier to sort and study.

How can this project help improve research methods?

Results may help researchers identify social, behavioral, and clinical factors in doctors’ notes.

What is the research team doing?

First, the research team is developing NLP methods to improve how the methods account for shortened phrases that clinicians may use in place of longer medical terms.

Second, the research team is developing methods to combine medical knowledge with statistical NLP methods. With these methods, NLP systems can use statistical techniques to manage phrases that occur often in the text and draw on medical knowledge for phrases that don’t occur often.

Third, the research team is creating a new NLP software package called SODA. SODA provides researchers with a way to use the new methods to identify social, behavioral, and clinical factors in EHR notes. It also organizes and stores the information so researchers can study these factors. The research team is linking SODA with existing NLP systems. At sites in Florida and New York City, the team is testing SODA to see how well SODA sorts cancer risks and patient traits in lung cancer screening.

Research methods at a glance