In order to introduce new technologies into the healthcare system, their safety, robustness and reliability must be tested in detail. The European Union has high quality requirements for AI and robotics in particular, but there is still an inadequate test infrastructure for developing standards, testing innovations and certifying new products.
The Fraunhofer Heinrich Hertz Institute (HHI) is working on the EU project TEF-Health (Testing and Experimentation Facility for Health AI and Robotics) to establish a test infrastructure for artificial intelligence (AI) and robotics in the healthcare sector. The TEF-Health consortium aims to improve this process and accelerate the validation and certification of AI and robotics in medical devices. To this end, the team is developing a (virtual and physical) test infrastructure that can evaluate various technologies in realistic environments, including hospitals and laboratories, for example. There, for example, software for patient care or diagnostics as well as surgical or nursing robots can be tested.
In order to develop this test infrastructure, the project partners are working on new regulatory and ethical requirements such as standardized test protocols and certifications for AI-based technologies or a specific code of conduct for their use. In addition, the necessary technical and administrative procedures need to be developed. In addition to leading hospitals, universities and clinical research institutions, government testing organizations such as TÜV, the German Physikalisch-Technische Bundesanstalt and its French counterpart, the "Laboratoire national de métrologie et d'essais", with which Fraunhofer HHI is working closely on this project, are therefore also on board with TEF-Health.
Fraunhofer HHI is involved in TEF-Health with its research groups "Explainable Artificial Intelligence" and "Applied Machine Learning" from the Department of Artificial Intelligence. Methods and instruments for the quality assurance of AI technologies are being developed here. In addition, the research team will be involved in the design and development of the virtual test infrastructure, contributing its expertise in areas such as time series analysis, the application of explainable AI methods and standardization.
The Fraunhofer HHI team has already gained experience in the field of AI standardization at national and international level. Since 2021, they have been organizing the annual workshop series "Towards Auditable AI Systems" together with the TÜV Association and the German Federal Ministry for Information Security (BSI), which produces groundbreaking white papers. Internationally, Fraunhofer HHI researchers are working on the development and standardization of AI procedures for the healthcare sector in the ITU/WHO focus group "AI for Health", which is led by the institute.
The evaluation resources and infrastructure developed in the project will be available to the industry in the form of fee-based services. Through these comprehensive testing and evaluation tools, the TEF Health team also aims to increase public confidence in innovative AI and robotics technologies.
Source: Fraunhofer Heinrich Hertz Institute
