Algorithms in Search of Anomalies
The project initially focused on the algorithms for automatically reading in and intelligently evaluating the mountains of paper. This is because almost every care service has its own paper documents with different structures, which are often only available in analogue form and even contain handwritten elements. The automated checking of these posed a real challenge to those involved. »These different documents are manually transferred into tables and checked. This is tedious diligence. With image processing, we speed up this process,« explains Dr. Henrike Stephani, deputy head of the department »Image Processing« at Fraunhofer ITWM. »Both the document structure can be captured with our algorithms as well as the contents. For example, they find signatures and assign them to the right people.«
At the same time, tour and duty schedules are the planning basis of care services. Comparing these with the billing data can provide clues to cases of fraud.
»One conspicuous feature, for example, is that many services are billed, but the tour plan only lists a short assignment. We have to find such peculiarities automatically,« adds Dr. Elisabeth Leoff. She is project manager and deputy head of the department »Financial Mathematics« at Fraunhofer ITWM. »Of course, our AI will not replace humans in the process in the future, but it will support them considerably.«