NLP Solutions Can Drive Payer Efficiencies – But They Are Not All Equal

By Kyle Silvestro, CEO of SyTrue

Payers are starting to move their outsourced payment integrity programs in-house in an effort to reduce spend, improve cost-effectiveness and control results. Clinical Natural Language Processing (NLP) solutions can drive these efforts forward – but not all solutions are built the same. Depending on your goals, bandwidth, expertise and budget, choosing the right solution can be the difference between a long-term successful program, or a costly setback.

As organizations seek to improve finances and operational efficiencies, payers are realizing the impact of unstructured data and the potential that NLP solutions offer. According to Market Insight Reports, the global market for NLP in healthcare and life sciences was valued approximately at $1.5 billion in 2020 and is anticipated to increase at a compound annual growth rate of 19% between 2021 and 2027.

By giving computers the ability to read, understand and interpret clinical language, NLP can extract and organize data from an individual’s episodic health record. By strategically presenting these data points, organizations can modernize the chart review process and eliminate antiquated and bloated workflows associated with manually reading a medical record.

Additionally, NLP gives organizations the power to retrospectively analyze longitudinal health data to find even one particular piece of clinical information about a single patient or identify subsets within populations that require further exploration.

Implementing NLP solutions can cause more harm than good if you don’t know what to look for. Missteps can be costly, time consuming and result in reduced productivity and setbacks, which is why payers evaluating their clinical NLP options should include these key considerations in their selection process: 

Moving Claims Review In-House

  • Claim selection – On average, 70-80% of claims that are reviewed have no findings that lead to payment recovery for the payer. An NLP solution can help you identify the 20-30% of claims where a discrepancy is apparent so reviewers are better focused on charts that make a clinical and financial impact.
  • Platform analytics – When it comes to payment integrity solutions, understanding how your reviewers are performing is key to measuring success. Analytics measuring performance at the auditor level, department level, and across all users will help you identify review types that could benefit from a more precise rule framework.
  • Time to implement – Understanding your project goals and timeline will help you narrow down an NLP vendor who can meet those demands. Projects and new endeavors have some fluidity, an NLP vendor should work with you to generate a project plan which can be delivered in your time frame.
  • Cloud infrastructure and hosting – Determine if the vendor offers flexibility on how and where clinical data is hosted. Some organizations opt to deploy in their own infrastructure while others prefer to use their vendor’s infrastructure.
  • Volume and scalability – Medical records can be tens of thousands of pages long. Because it is expensive to have your clinical reviewers waste time waiting for documents to process, it is critical to look for a solution that scales with your document volume requirements.
  • Security / credentialing – With solutions that offer single sign-on, users are not burdened with yet another set of login credentials to access the NLP system, in addition to logins for their claims administration systems, audit review platforms, or other programs. Data security is a top priority when processing all medical documents.
  • On-demand training capabilities within the platform – Different users access the system for different reasons: clinical coders or nurses for chart reviews, data scientists for rule building and customization, etc. Consider solutions that offer on-demand training that is customized by role, making it easier to onboard new team members. 

Achieving Higher Auditor Efficiency 

  • Clinical context – It is imperative that an NLP system accurately interprets context when analyzing data. For example, “AMI” documented in a psychology medication list (amitriptyline) and “AMI” documented in an emergency room intake summary (acute myocardial infarction) are two very different things. Your NLP must have the ability to consider the record type, document type and section that an extraction is coming from to avoid confusion and errors.
  • Building trust in extractions – Clinical reviewers are accustomed to viewing medical documents and scrolling through thousands of pages of notes. When all their data is accessible in a neatly organized view, users may feel it is too good to be true. Reviewers need the ability to not only see their data but also have the option to validate findings through an effortless user interface.
  • Communication and collaboration – All payment integrity reviews are a collaborative effort where different levels of expertise and/or multiple review passes are required. An NLP solution should maintain a history of each user’s activities and notes, auto-tagged with their credentials, to preserve a chain of custody and support a collaborative environment.

Leverage Experts While Investing in Your Own Growth

  • Domain expertise  Few payers have the financial and technical resources to customize off-the-shelf NLP solutions to fit the complex needs of their organizations. By selecting a vendor with deep domain expertise that offers relevant use-case-ready solutions and the services of data scientists and clinical experts, payers can more rapidly gain efficiencies from their NLP solution and realize a positive return on investment. 
  • Accuracy and precision  Some NLP systems provide precision scoring, which is an indication of the system’s confidence in the accuracy of a data extraction. Precision scoring can provide end users with assurances about the integrity of the data they are analyzing. For example, a designated precision score for a certain project gives a user a good understanding of the likelihood that extracted data is free of false positives and false negatives. Buyers should look for NLP solutions that consistently deliver higher precision scores.
  • Document annotation – Your vendor should employ clinically trained annotators dedicated to analyzing document extractions and accurately correcting false positives and false negatives. This method helps to train your NLP rules within a clinical context. 
  • Customizable authoring tools – An NLP system should allow data scientists of all skill levels to build, persist and share data assets including rules, algorithms, and methodologies. The ability to customize rules for extractions allows your organization to grow and onboard new audit types as needed without a heavy lift from your vendor.It also needs to be flexible so it can provide a broad range of general clinical extractions or findings for very complex clinical audit protocols that have very specific rules. 
  • Invest in your growth  Clinical NLP is an engine that drives many different solutions. Seek a solution provider who understands your business needs. An NLP vendor should not only empower you with tools, but also be an educator and sounding board for future growth initiatives.

Clinical NLP can be a powerful tool for the efficient management and interpretation of today’s overwhelming volume of healthcare data. By partnering with the right vendor, payers can overcome the challenges of dirty, incomplete, and inaccessible data and advance their business and clinical goals. 

Shares
Skip to content