Business challenge

In Germany, the state reimburses civil servants for certain healthcare costs. How could the Landesverwaltungsamt Berlin – which manages payments in the Berlin area – crack down on fraudulent claims?


With an IBM Watson® Fraud Cockpit built on IBM Watson Analytics, the Landesverwaltungsamt Berlin can pinpoint potential cases of fraud among millions of legitimate claims for reimbursement.



suspicious incidents among millions of legitimate claims


avoid misuse of public money


massive time savings by enabling targeted investigations into potential fraud

Business challenge story

Special support for public-sector personnel

In Germany, public-sector employees such as teachers and civil servants are entitled to special reimbursements on certain healthcare expenses under a system known as Beihilfe. After receiving treatment, the patient pays the doctor or pharmacist, then claims the expenses from the local authority – which for eligible residents of the Berlin state is the Landesverwaltungsamt Berlin.

In the vast majority of cases, the Beihilfe system swiftly compensates eligible people in the state of Berlin for their healthcare expenses. In doing so, the Berlin state provides a duty of care for its public-sector employees.


A small minority of people abuse the system and submit fraudulent claims and bills for reimbursement – for example, claiming for treatment that they did not receive, or altering the figures on a doctor’s legitimate invoice.

Sascha Nali, Head of the Central Beihilfe unit at the Landesverwaltungsamt Berlin, begins: “To pinpoint fraud and misuse of the Beihilfe system, we need to analyze data from millions of claims. State-of-the-art technology from IBM Watson identifies unusual claims and potential fraud, to help prevent public money being wrongly paid to fraudsters.

“In the past, we performed these investigations manually. It was very time-consuming to find indications that a claim might be fraudulent. For example, checking whether the patient had already claimed for the same procedure a few months previously was a very time- and labor-intensive process. Because it was hard to detect patterns in claims, fraudulent ones often slipped through the net.”

Aiming to accurately pinpoint fraudulent claims among millions of legitimate ones, the organization set out to find a more efficient and effective way of working.

Using cognitive analytics from IBM, we are clamping down on fraudulent claims and preventing unjust payouts – saving public money.

Sascha Nali, Head of the Central Beihilfe unit, Landesverwaltungsamt Berlin

Transformation story

Finding a needle in a haystack

The Landesverwaltungsamt Berlin enlisted IBM for support as it overhauled its approach to fraud detection. It decided to build a cognitive solution for fraud detection based on IBM SPSS® Modeler and IBM Watson Analytics software, which uses cognitive algorithms to refine the results of its analyses.

First, the organization used IBM SPSS Modeler to analyze over ten million claims submitted over a four-year period. The data was anonymized to protect the individuals’ identities.

The organization developed an algorithm in SPSS to identify cases with similar patterns in claims – for example, looking at the individuals’ claims history, and checking the claim against the average fee for the same treatment. Using this data, the Landesverwaltungsamt Berlin scored claims on how likely they were to be fraudulent, and compiled a shortlist of the top 100 people that seemed most likely to have submitted inaccurate claims.

Next, the Landesverwaltungsamt Berlin performed further analysis on the shortlist using IBM Watson Analytics, which enables staff to visualize the data. By using the predictive functionalities in IBM SPSS Modeler and IBM Watson Analytics, employees can identify cases with similar patterns in claims, which could indicate potential fraud. For example, employees processing claims can analyze each claimant’s history to identify outlier claims or unusual activity, which provide a good starting point for investigations. In total, the Landesverwaltungsamt Berlin analyzed claims worth EUR15 million using Watson Analytics.

As employees proceed with claims investigations, they can adapt the risk score given to each claim or patient based on their findings, or identify confirmed cases of fraud. Using its cognitive capabilities, Watson Analytics takes advantage of this information to refine its algorithms, becoming smarter at identifying cases of fraud.

Results story

Saving public money

With the new solution in place, the Landesverwaltungsamt Berlin is improving the effectiveness and efficiency of its fraud detection efforts.

Sascha Nali continues: “The IBM Analytics solution has already enabled us to prove some cases of fraud, so that we could reclaim funds. We’ve also highlighted many other suspicious cases that merit further investigation. By accurately pinpointing a handful of cases of fraud among millions of legitimate claims, we can help avoid misuse of public money.

“What’s more, we save time on our investigations because we can target our efforts much more accurately towards cases that require greater scrutiny.”

The Landesverwaltungsamt Berlin is in close contact with its counterparts in other states across Germany, so that they can work together to improve fraud detection. By combining their efforts, the organizations will be able to achieve economies of scale and cost savings. In addition, by pooling their claims data, the organizations will generate a much larger data-set to refine their fraud-detection algorithm.

Sascha Nali concludes: “Using cognitive analytics from IBM, we are clamping down on fraudulent claims and preventing unjust payouts – saving public money.”

Landesverwaltungsamt Berlin

The Landesverwaltungsamt Berlin acts as an internal service provider for personnel in public-sector organizations across the state of Berlin. Besides managing the Beihilfe unit, among its broad array of duties, the organization manages around 16,800 personnel files, including distributing around 17,000 child welfare payments every month for 41 local government bodies and providing support for around 65,000 care recipients.

Solution components

  • Cognos Analytics
  • Govt: Government Accountability - Cost Efficiency
  • Govt: Government, State/Provincial/Local Transformation
  • SPSS Modeler
  • Watson Studio & Modeler

Take the next step

IBM Analytics offers one of the world's deepest and broadest analytics platform, domain and industry solutions that deliver new value to businesses, governments and individuals. For more information about how IBM Analytics helps to transform industries and professions with data, visit Follow us on Twitter at @IBMAnalytics, on our blog at and join the conversation #IBMAnalytics.

View more client stories or learn more about IBM Analytics