Analyzing Large Data Sets Becomes Critical as Customer Base Grows
Accessing and analyzing “Big Data” plays an important role for economic improvements in healthcare. GWQ ServicePlus thus seeks to continually enhance its database and information management system to open new market opportunities. Having improved database performance and report generation is also crucial as GWQ increases its customer base, which leads to the need for a larger product set.
One of the key services GWQ provides is detecting indications of fraudulent invoicing through data analysis. By recognizing and responding to this need, GWQ helps health insurers answer key questions relating to profitability and quality assurance.
The data that GWQ tracks through services grows quickly because of the amount of activity that occurs as healthcare providers treat patients. Looking for inconsistencies and possible fraudulent invoicing thus becomes more challenging.
For example, over the course of several months, a physician might prescribe medicine multiple times for a single patient. Or several doctors might treat a patient through referrals. Both scenarios need to be monitored closely since medical costs may be inappropriately increased by inconsistencies hidden within the databases of health insurers. Whether such activity occurs accidentally or intentionally, experts estimate that fraudulent invoicing causes damages worth several billion Euros in the German healthcare sector every year.
The increasing number of customers that GWQ provides services to for tracking this type of activity has also contributed to the rapid growth of the data that GWQ compiles related to doctors, insurers, treatments and prescriptions. The more data that GWQ collects, the more precise the analysis needs to be to avoid errors that can cause insurance company customers to question the validity of the results.
Column-Oriented Database Reduces Time to Generate Reports
The relational database that GWQ previously relied on—Microsoft SQL—had reached its limitations in handling all the data GWQ had collected due to the increased customer base. GWQ thus decided to switch to Sybase IQ as a column-oriented analysis database. Sybase IQ, combined with RayQ Power data mining software, gives GWQ significantly higher report-generation capabilities for the analysis of the healthcare data generated by health insurers.
The column-oriented database technology of Sybase IQ enables quick analysis of massive data sets in contrast to row-based database solutions—where each data entry is read separately so that individual queries only accesses the data relevant to that specific query. At the same time, Sybase IQ does not require the time-consuming definition and maintenance tasks that are typical for supporting conventional SQL database queries.
For quick analysis of health data, RayQ Power writes query results as views into the highly-compressed data in Sybase IQ. This eliminates the compression and de-compression processes that are usually necessary for the exchange of data between the database and the analytics tool—and which delays the report-generation process. The speed of Sybase IQ thus directly benefits end users.
Multi-Step Plausibility Checks Cleanse Heterogeneous Data from Multiple Sources
“From the beginning, we were able to adapt Sybase IQ to meet our requirements, and the performance benefits of the new analytical data warehouse have since positively impacted the overall performance of company,” says Robert Raschka, Data Warehouse Manager at GWQ. “With the new system, we can prepare for further customer growth and simplify the development of new products.”
To deploy the solution, GWQ partnered with RayQ supplier Qyte, which helped design the solution and managed the project implementation, timeline and budget. The flexibility of Sybase IQ played an important role in the success of the solution because the GWQ requirements resulted in a number of adaptations to the data warehouse.
The data originates from a central data pool provided by the National Association of Statutory Health Insurance Funds, which receives data from health insurance companies in Germany. The challenge in analyzing the data pool arises from the diversity of systems and data-transfer channels used by the various health insurance companies that result in heterogeneous data.
To address the situation, Qyte developed check and plausibility routines to cleanse the data. Experts from Sybase and Qyte compiled professionally-founded standards of quality to evaluate incoming data. The result was a multi-step processes for the import of health data into the new Sybase IQ data warehouse.
“Currently, 16 multi-step check routines run on 450 data objects during data import,” says Raschka. “In addition to the data cleansing, invalid data is filtered out to enable us to check inconsistencies for their possible causes.”
New Analysis Capabilities Uncover Fraudulent Account Activity
Analyses have repeatedly uncovered incorrect physician medical accounts. Having this capability is important since the checking of such accounts is one of the many tasks that health insurers are legally required to do. Statutes in Germany regulate profitability and accounting audits as well as the persecution of misbehavior.
After GWQ informs a client about a faulty account, the client turns to the local National Association of Statutory Health Insurance Funds. The association contacts the responsible physician to correct the mistake and prevent it from happening again. Accounting checks by GWQ sometimes even uncover health examinations that are not necessary and were thus fraudulently billed based on the patient’s condition.
Automatic Data-Archiving Enables Recording of Deleted and Overwritten Files
After cleansing and validating the imported data, the next big challenge was versioning incoming data deliveries. From time-to-time, health insurance companies find mistakes in their data due to their own validity checks that they must communicate to the national association so the incorrect data can be recalled. The corrected data set version is then resent.
Before implementing the Sybase IQ solution, GWQ could not track which data sets had been exchanged, corrected or deleted during that process since the company could not access recalled data. Only the view of currently relevant data in the association system was possible. But this limitation changed with the multi-step versioning and the archiving of all data extracts that the new Sybase IQ data warehouse enabled. The archiving algorithm ensures that each monthly data extract from the association is matched with the live data in Sybase IQ. The result is a data warehouse that records all data that is deleted, overwritten or added by the association.
“This automatic data match enables computing-intense comparisons, which is one of the primary reasons for which the Sybase IQ data warehouse was designed,” Raschka says. “Due to the archiving, those data amounts are still growing with each data import, which the deployed data warehouse should handle without restrictions. The performance possibilities of Sybase IQ are not yet exhausted—the solution could easily process 500 billion data sets in a pharmaceutical package tracking project.”
Cleansed Data Pools Enable Improved Service and Increased Profitability
In addition to the archiving, the new solution also allows mapping of insured individuals and insurance data. Mapping simplifies the overview in case an insured individual switches insurance plans and is also a pre-requirement for a large number of GWQ services such as benchmarking analyses.
The management of data imports along with the archiving and data mapping within the new analytical data warehouse gives GWQ a clean data pool as a basis for its services. This supports health insurance companies in their high-quality provisioning of service to insured individuals and at the same time enables a high-level of profitability analysis. For example, with the data analysis it has generated, GWQ has been able to help its clients negotiate discount agreements with manufacturers of medicinal products. This has generated triple-digit million Euros in profitability reserves per year.
With doctor-based care, analysts can also manage doctor contracts more effectively. Results of account discrepancies and cleansing calculations are determined automatically and are made available for health insurance company customers.
Reduced Analysis Runtimes Enable Report Generation for More Clients
“The run times of our analyses have been reduced considerably,” Raschka says. “For some checks, we now need only a few hours instead of days. This improves our service and gives us further capacity to serve more customers with demanding analyses at the same time.”
According to Raschka, such demanding analysis reports had to be run one-at-a-time with the Microsoft SQL database, but Sybase IQ generates analyses in parallel. Furthermore, Sybase IQ enables GWQ to expand existing service offerings and simplifies the development of new products for which GWQ depends on data analyses.
For GWQ analysts, it is an important advantage to be able to model their analyses on their own—without any SQL knowledge. “Staff training only required two or three days,” Raschka adds. “Afterwards, our staff was in position to explore the comprehensive data pools for variable criteria.”
The quick analysis of massive data sets—made possible by the new Sybase IQ data warehouse—opens new expansion possibilities for GWQ. In addition, the company’s customers utilize the warehouse to uncover financially-relevant mistakes in health data and background processes. This demonstrates how the analysis of “Big Data” can help leverage potential economic improvements in the healthcare industry.
“Managing and analyzing massive data sets is a challenge with traditional systems that have reached their limitations,” Raschka says. “But by employing Sybase IQ with RayQ Power, we know we will not reach our limitations for a long time.”