Skip to main content

Using Clinical Natural Language Processing Solutions to Solve Payer Challenges, Improve Outcomes

May 18, 2021

Frank HeadshotBy Edan Stanley

As a tumultuous year for health care continues to expose challenges and high costs across the industry, payers are turning to technology more than ever to simplify processes and produce better outcomes for members and patients. Wolters Kluwer’s new Chart Review Accelerator software is designed to scan records and identify clinically relevant information such as conditions, medications, procedures, allergies, and lab results so that payers and providers can identify gaps in care, improve outcomes, and lower costs. It supports risk adjustment, quality measure reporting with more accurate data, faster medical chart review to facilitate pre-authorizations, and predictive analytics, the company said in a press release.

Frank Jackson, vice president and general manager of health language, Wolters Kluwer, offers insight on the role of clinical natural language processing (NLP) solutions to address payer challenges, and explains how this technology will support more robust and effective population health efforts.

Can you tell me about Wolters Kluwer Health's latest announcement regarding the Chart Review Accelerator and the Health Language platform? What challenges led your team to utilize this technology?

First, a little history. Health Language has been around 15 to 20 years. Originally, we built software that was embedded in electronic medical records (EMRs) to help physicians document conditions within the EMR. We, as an organization, have a lot of knowledge in how physicians document, including the shorthand, jargon, and correlating said jargon to the various ICD‑10 codes. We have an extensive background in understanding how providers navigate those systems and data.

We have realized that there is a goldmine of insights in the unstructured part of charts and EMR. With that in mind, we combined the clinical lexicon of 15 years of providers’ synonyms and how they document and the emergence of artificial intelligence (AI), to try to attack a problem for both payers and providers—unlocking that unstructured data and putting it to good use.

We have seen that payers and providers are thirsting for new insights to solve and improve patient outcomes. We estimate that the unstructured portion contains 80% important data, which was not being utilized. We arrived at it through a clinical perspective, and then we added machine learning and technology.

A challenge with integrating NLP technology and health care is that there is a lot of associated hype—just having machine learning was not good enough. We needed to be good at machine learning. We needed to be at an advanced capability. We built a capability in our engine, so we can tune it for specific accuracy and precision based on the use case of the payer or provider.

What about this technology is uniquely beneficial for clinicians and what about for payers?

The answers are a little different, but there is some overlap. On the provider side, EMR technologies were not designed to be able to extract insights at the population level. It was a tool meant to document the visit, so extracting the correct data was difficult. A lot of it was stuck in unstructured data, in the progress notes.

What we have done for our new release with the Chart Review Accelerator is add a clinician‑friendly workflow in which they can integrate to solve their problem. There is a variety of use cases, but workflow is what we added specifically for providers.

Then on the payer side, we have intentionally prebuilt models to specifically solve some of their problems. Unique payer challenges tend to stem from analytics, closing care gaps, finding hierarchical condition categories (HCC) in support of risk adjustment, and developing a more accurate 360 view of their members.

A lot of payers have built in analytic functions, and they have data science. We have prebuilt these analytic application programing interfaces (API) to extract insights but then have it feed potentially like a predictive model that supports underwriting or other functions within the payer platform.

What role will the Health Language Chart Review Accelerator play in improving patient outcomes?

We think about that a lot, because if the technology does not improve outcomes, then we are not hitting the mark. The answer to that is in thinking about time post‑COVID. The last year has been a very unusual world for both payers and providers.

Obviously, the pandemic changed a lot of people's lives, and the health pattern has changed quite a bit. People did not have elective surgeries or a maintain regular doctor visits. There were many forgone aspects of patient care.

In the payer world, for example. they generally try to look at last year to figure out what they are going to do this year. However, looking at last year with some of their same algorithms was not a good proxy for their plan. Chances are, with the pandemic, records or claims around some of the data points that payers needed to predict what members needed, such as more interventions around disease management, were unavailable.

The Chart Review Accelerator is designed to take a deeper dive into all corners of the chart, go back more than the past year and improve the process of finding real insights around disease and health for a member to better inform payers and providers going forward.

Will utilizing this technology have any effect on cost or reimbursement?

We believe so. The early indication for the market is they agree. One of the benefits to cost is if there is an early identification with clinical granularity around health conditions, emerging conditions, or the severity of a particular condition.

If a provider can find and deliver that at the right time, they can proactively schedule clinical interventions to reduce cost, avoid a hospitalization, or avoid a complexity in somebody's health that would drive up cost.

Then, within the payer there is a much more cooperative attitude, driven by the trend around payer and provider collaborations, value‑based agreements, and accountable care organizations. As payers find insights, they are welcomed by providers who are working to improve cost efficiencies.

We are seeing and targeting risk adjustment in the payer world. More accurate identification of health conditions improves their risk adjustment reimbursement. We are also getting traction in the market around accuracy in HCC.

Is there anything I have not asked you about or anything that you would like to add?

I would add that the medical chart is a goldmine of data. It is locked in this unstructured data. Payers have exceeded value from the claim, and they're thirsting for more clinical insights. The answer, we believe, is that goldmine locked into the chart.

This Chart Review Accelerator, we believe, is a solution. Our intent over time, and there is a wide application of this, is that we would love to be the pervasive intel inside for chart reviews within the health care ethical system. That would be our goal.

About Frank Jackson

Frank Jackson has been in the payer, managed care, and provider world for about 25 years. He specializes in managed care, having spent most of his time working with organizations like Blue Cross and Kaiser Permanente.

Mr Jackson’s expertise is ranges from Medicare to Medicaid to commercial insurance. His core background experience comes from his time working with organizations to create actionable insights to improve health care patient outcomes. In his current role with Wolters Kluwer, Mr Jackson navigates the intersection of health care innovation and technology to deliver overall improvement to the care process and better patient outcomes.

Agree or disagree with an article? Share your professional thoughts on an article you read.

Your Name
3 + 1 =
Solve this simple math problem and enter the result. E.g. for 1+3, enter 4.
Back to Top