As the healthcare industry shifts to outcomes-based financial models, coupled with requirements to improve the member experience of healthcare, plans are turning to robust data analytics (typically found in consumer-driven companies like Amazon or Verizon) to better understand their members as healthcare customers. Alongside the well-structured claims data they have about member encounters, health plans are looking to incorporate non-traditional data from self-generated periodic surveys, consumer reporting agencies, care management notes, and clinical data (both structured and unstructured) — just to name a few.
"Using Natural Language Processing (NLP) technologies to parse through the sea of unstructured clinical, care management, and customer service notes can create new insights about members and populations."
Fold in social determinants of health (e.g., economic stability, neighborhood and physical environment, healthy food availability, etc.), and plans have the ability to know their members and their customers in a much more meaningful—and potentially actionable – way. All of this data positions data scientists and analysts to develop richer member profiles using innovative data mining techniques. This, in turn, arms decision makers with data they can use to develop novel strategies and programs for members. Member profiling allows plans to target programs and resources where they will have the most impact, rather than a shotgun approach to cast the widest net and see what works.
Plans are also using data and predictive analytics to develop interventions that help to improve patient outcomes. One plan, for example, created a model to predict an elderly member's risk for falling using structured historical claims data, social determinants (Are they living alone? Are there stairs in their home? Do they have access to nutritious food?), and survey data that assessed the member’s own perception about their risk of falling. They then created a targeted program to provide these members with a PERS device (Personal Emergency Response System – think MedAlert) they could use to contact caregivers if they felt unsteady or alert medical personnel if they fell. Creating this early invention program helped avert more serious injuries and costly hospitalizations.
As plans begin to receive clinical notes from hospitals and providers through the implementation of Health Information Exchanges (HIEs), they are starting to comb through previously untapped resources to identify whether certain services are being performed or trends in care. Natural Language Processing (NLP) unlocks a key strategy in tapping into unstructured patient clinical notes. In fact, during the HEDIS (Health Effectiveness Data Information Set administered by NCQA) season, plans are starting to use clinical notes to reduce the burden of the dreaded annual ‘chase’ process, where legions of staff are sent from the plan out to provider offices to examine and retrieve medical records to supplement their HEDIS reporting. This is an ongoing point of friction at providers’ offices, where they need to provide these records to nearly all of the health plans they work with.
As plans continue to enrich their data stores and use advanced data analytics, they are able to help facilitate the right interaction, with the right patient, at the right time, while reducing provider friction. Ultimately, this is a recipe that focuses on what’s most important—achieving the best possible outcomes for their members.
You Might Also Like:
For Payers, a Clinical Data Strategy is Critical
Opportunity for Payers – There’s clinical data out there that you may need, but don’t know it
About the Author
Mark Taylor is the Director of Strategic Consulting at Ready Computing in NYC. He has deep experience working for highly rated health plans leading administrative system and quality improvement strategy implementations. He is passionate about adapting Health Information Exchange technology in the payer space to integrate real-time data and improve patient outcomes. mark.taylor@readycomputing.com
