The prevailing global economic conditions are negatively impacting the healthcare industry. Continuing staff shortages, endemic COVID-19, high inflation rates, and supply chain disruptions continue to drive medical costs higher, worsening what is already a persistent challenge for health insurers.
To offset rising healthcare costs, health insurance payors are challenged to find ways to contain their own costs by streamlining their processes and tightening their belts, which invariably sees their handling of unstructured data come under increased scrutiny.
Health insurance payors handle large volumes of unstructured data from healthcare providers, vendors, policyholders, and others as part of the revenue management cycle. These unstructured documents, many of which are handwritten, require manual processing and data extraction, which is extremely time-consuming and dramatically increases administrative costs.
In a largely paper-bound industry like healthcare, automated intelligent document processing opens up multiple opportunities for cost savings through streamlined workflows.
Below are six common ways in which health payors can use intelligent document processing to rein in medical costs, make their internal processes more efficient, and uncover deeper data-driven insights.
- Appeals and Denials
Providers can appeal a payor’s decision to deny payment for a submitted claim – demonstrating why the claim should in fact have been paid. These appeals are typically emailed to payors along with a variety of forms and attachments, some of which may be handwritten, as proof of the medical need.
The unstructured documents involved in the appeals process are highly variable and typically handled via manual processing and review. The time involved in manual reviews in turn delays appeal decisions, potentially damaging payor/provider relationships. At the same time, the labor costs involved are high as payors may have a backlog of tens of thousands of appeals in process at any point in time.
Intelligent document processing leverages deep learning technology to ingest, classify, and extract relevant data from these highly variable documents that we know as unstructured data. By automating what was once a tedious, time-intensive process, intelligent document processing shortens the appeals cycle, which reduces operational costs and facilitates better payor/provider relationships.
Some insurers require preauthorization for specific types of care. A prior authorization is a directive or request from a medical professional to preapprove a prescription or certain type of procedure or treatment for an insured patient.
Prior authorization requests are typically sent via fax or physical mail. They often include handwritten notes from providers, patient medical histories, clinical notes, and other highly variable documents. As with appeals, prior authorizations are often manually processed because of the variability and unstructured nature of the documents provided.
Intelligent automation uses deep learning to pull critical data from this unstructured data, organize it, evaluate it, and learn patterns from it. As document processing accuracy and speed increase, the time from initial request to authorization is reduced, which helps patients get the care they need faster and reduces operational costs for the payor.
Onboarding a new provider and or his or her practice into an insurance network and creating a profile for inclusion in member network directories involves a lot of unstructured data. The information supplied by the provider and the practice or facility, will include the provider’s name, address, and other contact details which of course vary greatly from other providers’ applications and so require manual processing. Ideally, onboarding should be a seamless process, so that providers’ profiles become accessible to patients within the insurer’s network as quickly as possible.
Today, the process can be automated. Intelligent processing expedites the timeline for registration completion and improves the accuracy of provider information in network databases and directories.
4. Medical Claims
While a medical claim is one of the most basic documents in the insurance process, manual processing makes the processing thereof anything but basic. Claims can come in the form of invoices, letters, handwritten notes, and other highly variable document formats. Manual processing lengthens the claims cycle and slows payout times to providers or patients.
Intelligent processing on the other hand speeds up the timeline for claims review and improves accuracy in correctly matching procedure coding with covered services. Accurate claims decisions limit the number of legitimate appeals that payors have to process.
5. Commercial Risk Adjustment
One of the most critical areas for intelligent document processing is commercial risk adjustment. This process takes place in Medicare, where the payor is the administrator of plans that are paid by the U.S. Government through CMS. Medicaid and Medicare account for 36.5% of the U.S. health insurance market.
Payors receive medical records from providers of all services. The payor then adjusts and pays the claim before requesting reimbursement from CMS. Each service is coded with standard medical coding, such as ICD-10 codes, which the payor uses to determine whether a diagnosis or procedure is covered by the subscriber’s plan. Accuracy in records reviews is essential to determine how much money the insurer is likely to get back from CMS. Errors can lead to either over- or under-payment by the payor.
Intelligent document processing increases the accuracy of records reviews and speeds up the process, decreasing the time between when the payor pays claims and when they in turn receive reimbursement from CMS.
6. COVID Certification
COVID vaccination cards are a good example of how new or variable document processing occurs over time, in addition to the conventional document processing already described, . These cards are usually handwritten and typically require a manual review. However, with intelligent processing, they are faster to process, and data is extracted more accurately.
At some point, COVID certifications will probably become obsolete. However, other limited duration documents like this come around periodically in the insurance sector, adding to the enduring document types already described. Manual reviews on these short-duration documents are especially challenging given that there are no long-established protocols for them. Deep learning technology allows for rapid learning for efficient document processing.
Fast and accurate insurance document processing helps the overall health system to cut costs, ensure timely treatment for patients, and spend less time handing appeals and service calls. Plus, there is the improved reputation that ensues for insurers in the provider network among subscribers.
About Tom McCann
Tom McCann is the Growth Strategy & Business Development at Instabase, a San Francisco-based startup that created the created the first automation platform for unstructured data, enabling organisations to drive transformation across manual processes by unlocking unstructured data with deep learning.