Scalability in Private Health Insurance Benefit Accounting

A look at the work carried out in private health insurance benefit accounting departments reveals a number of daily challenges. In general, there is a high backlog in the processing of benefit applications, which is now distributed more evenly over the course of the year than was the case in earlier years. This is due to the fact that customer behavior has changed. The reason: Apps have made it easier to submit documentation, such as receipts, which can be photographed and submitted immediately. Irrespective of this, however, it is still the case that customers in certain contract situations, e.g. in comprehensive insurance, collect receipts over the course of the year in order to be able to assess whether it even makes sense to submit them if they have contract agreements involving deductibles or premium refunds. This leads to an accumulation of benefit applications at the beginning of the following year, which always causes the backlog of work to increase sharply in the first quarter. The constant backlog in the processing of incoming submissions negatively impacts the level of service that customers expect - they want to see their submissions processed as quickly as possible.

But why is the processing backlog large right now? The reason for this is cost pressure in private health insurance sector. Deploying more highly qualified employees to process submissions can only occur to a limited degree, especially if additional revenue from new business prohibits it. Since a long qualification period is required before an employee can be deployed to benefit accounting, it is not possible to address the processing backlog during a period with a high workload. As a result, customers end up with lower service quality.

Dark processing as a method for managing backlog

In recent years, "dark processing" has emerged as a method for dealing with this backlog. Dark processing means that there are no administrators involved in processing the cases. Instead, this is performed by the company's computer systems via automated processes. This requires an automated reading of the submitted documents (or the delivery of the relevant document data) so that the tariff conditions can be applied to the service provider's invoice (doctor, dentist, pharmacy, physiotherapist, etc.) in a set of rules. Most of the time, dark processing is performed with "simple" cases.

So what exactly are "simple" cases? Simple cases are the submitted documents from which the data can be extracted beyond any doubt and where the rules according to which the invoice is reimbursed are not in the heads of the administrators, but instead are stored in the computer system's algorithms. The "simple" cases are therefore usually already fully processed without the administrator having to intervene.

Using dark processing even when it is a complex case

Why is it then that computer systems do not store the reimbursement rules for more complex cases?

To do this, it is necessary to understand why computer systems were introduced in private health insurance benefit accounting. Most computer systems still being used today were not designed for automated processing, but for supporting and easing the burden for administrators who had mastered the rules for reimbursement. At the time that these systems first emerged, it was not at all possible to recognize the contents of the submitted invoices. Yes, in fact, many of the systems originated in the 70s and 80s of the previous century.
In many cases, the systems were only used to record the invoice data entered by the administrators, and to receive their manually executed settlements and send them (mostly overnight) as a mass printout. They then automatically paid out the money to the customer by forwarding payment orders to the banks.

Calculation rules derive the reimbursement from the invoice amount

The knowledge required for reimbursement was therefore not stored in these systems; after all, the administrators had access to it. In simple cases, calculation rules were stored that could automatically derive the reimbursement amount from the invoice amount, provided that the reimbursement as such was checked by the administrator according to the rules stored in the general terms and conditions and the document's header. At the same time, these old computer systems are expert systems that are not designed to guide and provide knowledge to specialists so that they can make the right decisions.

As a result, new employees in the field of private insurance benefit accounting require a long period of familiarization with their respective computer system. This makes it difficult to scale the processing of the emerging workload in the benefit accounting system on short notice.

So how do you make the work in benefit accounting scalable?

The mental migration solution approach.

In contrast to a classic migration, in which data and algorithms are migrated from Technical System A to Technical System B, acquired knowledge, trained behaviors, and a stored wealth of experience must now be migrated from an inhomogeneous group of employees (System A) to a Technical System B.

In classic migration, the three letters ELT are usually used for the three steps of migration: E for extraction, L for load, and T for transform.

A detailed overview of the steps involved in the mental migration process:

Extract: a multi-faceted process

Extraction is a difficult process, as is reflected in several facets: here we should mention examples such as the willingness to disclose knowledge or the fear of no longer being a knowledge carrier. In addition, there are difficulties in transferring the knowledge. This is because it is rarely the task of administrators. There are problems in preparing the knowledge so that it can be transformed, i.e., converted into rules, and it remains uncertain as to whether all the knowledge could be extracted. All of the knowledge is distributed in varying portions among many, diverse individuals. Here are examples from the many complex facets that one may find in an attempt to identify knowledge. At this point, it is essential that the process be overseen by trained and thorough employees who can deal with these special situations. The knowledge carriers must be shown how their tasks will look in the future: Not the billing employee who takes on tasks such as calculations and research, but the insurance expert who is relieved of mindless tasks and promoted to become the customer's consultant.

Transform: passing on knowledge from a human to a machine

Transforming the extracted knowledge from Human System A into Technical System B is a task for experts who are skilled in transforming knowledge so that the receiving system can store and apply it. In our projects for implementing the in|sure Health Claims benefit accounting system, these are benefit modelers and process modelers. These modelers convert the extracted knowledge into a benefit model that maps the administrators' knowledge, even going beyond the written knowledge and written work instructions in the general terms and conditions. This enables the model to be automatically applied to the reimbursement claims to the greatest extent possible. If a fully automated implementation is not possible, the necessary data is determined by posing simple questions to administrators, which is then followed by automatic calculation. Working with the benefit modelers, the process modelers adapt the processes so that the implementation is highly efficient. The necessary skills for this are also communicated to our customers during the mental migration process. This way they are able to successfully migrate or integrate new knowledge and ideas into the system during the system's implementation.

Load: receiving human knowledge

This step presents challenges for the receiving system. The receiving system, in this case in|sure Health Claims, must be suitable for receiving and also applying the administrators' knowledge. It should be completely automated, if possible. The system must therefore be able to acquire a wealth of experience and also be able to retrieve it as a basis for the behaviors in which it has been trained. This allows for the right decisions to be made. In the vast majority of cases, this does not mean AI, but rather rule-based behaviors, as every decision's traceability must be indicated at all times. However, the use of AI can be supportive, such as when identifying whether a bill for an inpatient stay is from a wellness clinic or a regular hospital. A very easy task for an administrator, as they can look at the statement. However, it is much harder for a recognition process that is rule-based than for one that is AI-based. If the receiving system is able to almost fully receive the administrator's knowledge, then all the conditions for mental migration have been provided.

Mental migration paves the way to scalability

A successful mental migration paves the way for scalability in benefits accounting: benefits administrators are brought to where they need to be by asking questions without having to familiarize themselves with a complex system of tariffs. This makes it possible to productively deploy new benefits administrators in the shortest possible time, or to reassign employees from other areas during periods of high workload. Based on the knowledge that has been transferred, the benefit accounting system itself is able to perform the necessary calculations and automatically searches in the system. Combined with well-extracted data from the submitted documents, this is the best possible dark processing method.