Data is now a vital part of the healthcare industry, and structured data alone will not be sufficient for better member outcomes. However, much of it is still inaccessible to analysts and experts, as precious information hidden within electronic health records (EHRs). Natural Language Processing (NLP) technology can unearth the 80% of unstructured information stored within the electronic health record.
NLP provides deeper, vital insights into members
NLP can help us to bring valuable missing pieces of information to the surface to cover risk and monitor patient’s health more closely. Disease severity is just one of those areas - for example cancer stage, is critical for proper member management, but the information isn’t always available consistently in structured claims, but is included in the unstructured data. Claims data gives a lot of detailed structured information about the individual, such as their medications, diseases they're suffering from, as well as procedures and treatments that they've had, but that only provides part of the full member picture.
Imagine Ryan, a 76 years old patient suffering from diabetes. Since his wife’s passing he lives alone at home with two cats. His medical record includes structured information such as his prescribed medications, and the procedures and treatments that he’s had. However, critical information is missing from the picture. In text you will find, Ryan suffers from loneliness and isolation since his wife’s passing, that he often forgets to take his medications, drinks at least 2 units of red wine a day, and doesn’t take the time to eat a well-rounded healthy diet. Items such as social determinants, need to be taken into consideration as they are critical insights into patient/ member outcomes and allows providers to intervene before a serious event or detrimental decline in health takes place.
NLP reduces manual effort, saving precious time
One large health system wanted to study approximately 100,000 heart failure patients with pacemakers. These patients had generated about 34 million documents that needed to be analyzed. Using Linguamatics NLP platform, the entire analysis was completed in just three months, including installation and training, by two people with no previous NLP experience. The accuracy of the system was 95-99%. For reference, it was estimated by the source that a manual review of these documents, without NLP technology, would have taken approximately 35 years.
Another advantage for payers and health plans is that NLP can produce insights on risk adjustment. Here we can look into diagnosis and demographic data for member populations and provide a risk score that can be used to adjust reimbursement calculations for those members. For instance, a patient with complex comorbidities that needs higher-comprehensive care will be associated with a higher payment. It is also important for NLP to unearth evidence and validation of chronic conditions especially those that have systemic effects on the members such as diabetes or multiple sclerosis (MS), in order for proper payments to incur.
NLP is now a mature, accessible AI technology for payers
Artificial Intelligence and Augmented Intelligence (AI) technologies are now able to support and automate many processes that help to improve users’ understanding of risk and predictive models in healthcare. The acceptance of augmented technology workflows by the general public is growing too.
There’s a lot being said about AI and NLP at the moment, and about the automated way of analyzing text, images and tables. But it’s important to keep the real goal in mind - which isn’t to replace people who are doing manual work and automate everything. Our goal is to help people do their jobs more effectively by using AI to improve health outcomes.
Watch our webinar to learn about the 5 ways payers gain value with Natural Language Processing.
More examples of Return on Investment here.
Contact us to discuss your data-related challenges and see how NLP can help.