There’s been a lot of buzz about the promise of artificial intelligence (AI) in healthcare. We’re moving rapidly from hype to real-world use cases in which health leaders recognize that AI holds the potential to diagnose and treat disease, improve processes, and better manage underlying operational, financial, and patient health data through which they can innovate and maximize value.
In today’s environment, Medicare Advantage health plans in particular are balancing the realities of competing in the present while preparing for the future, a difficult line to walk for most leadership. The hype around advanced digital solutions such as AI only makes establishing a strategic focus more difficult.
For health plans, this means balancing the need to oversee and manage beneficiaries’ health and outcomes with efforts to improve quality measure scores such as Star Ratings. But accessing and making sense of volumes of member healthcare data has proven to be a mammoth task. That’s because much of the data that’s generated throughout the healthcare journey, including data found in electronic health records (EHRs), is unstructured. This unstructured information can be found in a variety of patient medical records such as physician notes, labs, and medication lists, as well as an organization’s own data warehouse.
Why is unstructured data valuable to health plans?
Unstructured data often contains clinical insights crucial to improving performance across the organization and accurately classifying beneficiaries. By extracting key insights from unstructured data, plans gain a full view of the patient and their interaction with the clinician—information that impacts quality of care and therefore reimbursement.
But extracting the data is time-consuming and complex, often requiring dedicated teams of clinicians to sift through the notes to identify relevant information. Clinical Natural Language Processing (cNLP)—a category of AI—speeds up that process, filters out information that isn’t relevant, and helps to normalize or transform the data.
Properly asses patient risk
It is important for health plans to identify and extract valuable insights from member health data, tests, or lab results in order to properly assess patient risk. For example, payers need to know whether their members with type 2 diabetes have had an A1C test in the past year. Such information is important to accurately determine the Hierarchical Condition Category (HCC) classification for each patient to ensure adequate reimbursement for their high-cost member populations.
Addressing gaps in patient care
Unstructured data is also key to improving performance as assessed by the Star Ratings program and other quality measures. By accessing, codifying, and understanding unstructured healthcare data using cNLP, plans can determine how many members have used prevention services such as breast or colorectal screening, or flu vaccinations. This gives plans a broader picture of the member’s health, so they can consider how to manage possible gaps in care, such as reaching out to at-risk members who may have declined screening and encouraging them to get screened.
Improve patient outcomes
With the relevant information made readily available using clinical NLP, plans can ask appropriate questions and use the information to take action in areas that can improve performance, enable them to properly classify patients, and help improve patient outcomes.
Once a plan has access to the value locked in unstructured data and can map findings to industry-standard codes, they can drive significant improvements in clinical quality (HEDIS, Star Ratings), risk score accuracy, and financial performance.
Seven steps health plans can take to leverage the value of their unstructured data:
1. Gather data on member health, not just healthcare
Identify the data sources that can provide a holistic picture of the member’s health status and inform the organization’s risk strategies and population health efforts. Consider traditional data sources such as claims, clinical, and administrative data as well as social determinants of health (SDoH) data, which is critical for this high-risk, complex population.
2. Design an enterprise-level data strategy
This should include the ability to extract and aggregate all data types, combining unstructured data, such as free-text physician notes, with structured data, such as lab values and medication dosages. Instead of jumping right to the predictive models (often based on inaccurate or incomplete data), do the fundamental work that starts with defining a common language. Look for opportunities to collaborate and share data with providers and other third parties.
3. Evaluate adoption of cNLP technologies
Whether you go it alone, adopt a new platform, or integrate cutting-edge technologies into existing systems, determine what’s best for your organization.
4. Align internally
Create organizational alignment on how the data should be used to support the plan’s strategic focus areas.
5. Prove it, then scale it
Launch pilots within specific functions so the use cases for cNLP are well defined and lead to value—for the member, the health plan, and its provider network.
6. Get the data ready for analytics
Data and analytics teams must have the capabilities—and the knowledge—to help the organization comply with regulations based on data accuracy, but also to better understand the health of their member population and improve data-sharing practices with providers.
7. Establish a solid data governance structure to
Ensure all functions are consistently managing the data, normalizing it, and mapping it to industry standards.
Industry-leading cNLP solution from Wolters Kluwer
The Health Language Clinical Natural Language Processing (cNLP) solution optimizes manual medical record review, automates the review of unstructured data, extracts clinically relevant data, and codifies extracted data to industry standards. Speak to an expert to learn more about to the Health Language cNLP Solution.
