Reducing Patient Risk with Natural Language Processing (NLP)

Making a Difference and Reducing Patient Risk: I2E Natural Language Processing (NLP)

A valuable part of a clinician’s training includes the effective identification and careful documentation of all the elements impacting a patient's well-being. Thorough documentation is essential to ensure accurate and timely clinical care. Although electronic health records (EHRs) hold many great opportunities to capture essential details in electronic form, patients could be at risk if all elements of their medical records are not compiled and analyzed. With 990.8 million reported visits to physician offices in 2015 [1], odds are that precious information could slip off the radar of even the most dutiful clinical staff. 

The importance of Nature AND Nurture in healthcare

For this reason, providers must adopt a successful population management strategy that considers all elements of a patient’s record, including the estimated 70% of the record that exists as unstructured notes. Structured data is excellent for documenting patient information that ensures a hospital runs effectively but not as efficient for capturing imperative clinical concerns during a patient’s 20-minute encounter with a physician. 

Although the concept of nature vs. nurture has been well-documented for centuries, providers are just now realizing the critical importance of social determinants:

  • Does a patient live alone?
  • Do they utilize a walking cane?
  • Are they on a fixed-income? 

Identifying before protecting: Using I2E to help vulnerable populations

Undoubtedly, EHRs contain a wealth of information to identify patients requiring special attention, such as those with:

  • Chronic disease
  • Financial vulnerability
  • Need of social services
  • High risk for suicide

The problem persists that this vital patient information often gets trapped in the unstructured data of the provider’s EHR.

Like the classic spaghetti sauce’s in there. But you need AI technology solutions like natural language processing (NLP) to get all the critical details out of the “jar.” 

Our NLP text mining platform can extract unstructured data about determinants of health, where they are lifestyle choices (such as smoking, drinking, and exercise) or social determinants (such as social support network, living location, or ambulatory status) from the unstructured data. Once extracted information can be easily integrated with existing infrastructure such as EHRs, Hadoop, analytics tools, and data warehouses, and utilized by physicians to determine patients requiring special attention.

If you would like to learn more how Linguamatics I2E can be utilized within healthcare organizations, please contact us to learn more or request a demo.

You can also find out more and download our Mining Unstructured Patient Data for Successful Population Health whitepaper here:


This blog is based on my original response published in the Thought Leaders Corner in the March 2018 issue of Population Health News.

[1] “National Ambulatory Medical Care Survey: 2015 State and National Summary Tables.” U.S. Department of Health and Human Services, Centers for Disease Control and Prevention and the National Center for Human Statistics. 

Learn more about Population Health Management and Analysis