September 2025 | Austria | Artificial Intelligence (AI)
Digital Transformation in Austrian Accident Insurance: Automation, AI, and EFEU
For more than 15 years now, the Austrian Workers’ Compensation Board (AUVA) has been using the standard product EFEU (Electronic Determination and Processing in Accident Insurance) for its benefit procedures in accident insurance. This system was developed jointly with the biggest Austrian accident insurance providers, namely the Social Insurance Institution for Public Employees, Railways and Mining (BVAEB) and the Social Insurance Institution for the Self-Employed (SVS) and is used by all of them.
EFEU consists of two complex systems:
- A workflow-based core benefit system for administrative processing
- Digital online services for the target groups of accident insurance
Since 2010, the workflow-based core system has enabled fully digital processing of insurance cases involving occupational or service-related accidents and occupational diseases—from the initial report and case creation to the payment of benefits.
For several years, we have offered web forms in our customer portal, including for electronic reporting of occupational accidents. As usage was initially low, we launched several initiatives in 2024 to increase acceptance of these electronic reporting forms. As part of the AUVA website relaunch, a new web form was introduced, which can be used by reporting entities (e.g., companies, educational institutions). This form can now be submitted not only with ID Austria signature, but also via a low-threshold email/TAN verification method.
In addition, the legal regulations on reporting obligations were updated in 2024. Since then, electronic reporting of occupational accidents has become the preferred and mandatory method.
Thanks to these measures, the share of electronic reports increased to over 75% within one year. This laid the groundwork for automatically creating cases and developing a rule-based system that enables partial automation in case processing.
Since late 2024, algorithmic decision-making has been used for automatic case creation. Incoming electronic accident reports and initial hospital reports are analysed and classified based on a rule set using ICD-10 codes, reported injuries, and data on sick leave and treatment. Cases are automatically categorized as “no further action required” (currently about 35% of reported cases). These cases are either immediately archived or, if temporary incapacity to work is reported, monitored for up to two months before being either assigned for manual processing or automatically archived.
As a result, only cases with significant injuries or extended incapacity to work are assigned to case handlers for manual processing, ensuring that expert assessment is focused where it is truly needed.
Since June 2025, accident reports have been automatically analysed using generative AI (LLM, GPT-4o). The system processes the submitted data—such as the description of how the accident occurred—and extracts key statistical information, including the work process, physical activity, deviation, cause of injury, object involved, and material used. These details are then saved to the insurance case based on predefined value lists.
In the initial rollout phase, the results are still presented to case handlers for review and completion. Any corrections made are continuously analysed to improve the AI’s rule set, with the goal of enabling fully automated statistical reporting in the future.
Since June 2024, we have also been using algorithmic decision-making to process expert fee invoices. Through our online assessment portal, invoices are submitted electronically to EFEU and automatically processed and paid. Currently, over 80% of expert fees related to accident assessments are handled this way. Only invoices that include special services or exceed standard rates are forwarded to case handlers for manual review. ●