Natural Language Processing and Health Care Quality

Health care data has never been more important. Consumers are starting to push for greater transparency, value-based care programs keep changing and there is an urgent need for affordable care. These factors and more, mean that health care organizations must use data in a way that is meaningful and that results in better health outcomes. But the industry is grappling with proliferating technology solutions, hundreds of standards, and limited knowledge of how technology can reduce inefficiencies.  

There is, however, a way to lessen the data obstacles to better health care: Natural Language Processing (NLP) can automate the transformation of text into discrete data, like FHIR, to reduce reporting burdens and enable reallocation of resources from medical record review abstractors to clinicians – closing a care gap. It can also provide comprehensive information to guide quality efforts. If an organization receives a limited set of diagnosis codes that don’t represent the full story, for example, NLP can pull in descriptive information from problem list descriptions and unstructured data fields to fill in the blanks 

Sounds great! What’s the catch?

Of course there are critical elements to consider when implementing NLP. For example, with the ability to translate data comes the necessity for providers and care teams to invest in resources (funds, time, effort) that mitigate risk by ensuring delivered information is valid. Note that the key word in that sentence is invest. Investing in NLP can compound investment benefits that come from reducing burden, regardless of the initial lift required. But the decision to use NLP should be weighed against strategic priorities, resource capacity and the need for performance improvement.  

How do I get started?

Workflow should guide integration of new technology. A structured approach is important here, as it is in any process improvement: 

· Reflect on your mission and identify your quality goals. 

· Perform an assessment to understand your data’s quality and completeness. 

· Prioritize data gaps that limit your quality measurement goals. 

· Identify a set of unstructured data elements that can boost your data’s integrity. 

· Pilot NLP technology by performing iterative data validation.  

· Create a plan for ongoing monitoring and improvement. 

With the evolution to digital quality systems, how are you leveraging technology to support delivery of the best care? How can you apply NLP to those efforts? Share your thoughts in the Community Forum.