AI in medical documentation - Legal frameworks and recommendations

Increasing administrative requirements are placing a considerable burden on healthcare professionals. The preparation of medical reports takes up a lot of time, with specialist staff often transcribing voice recordings manually and, in some cases, still writing reports by hand.

Artificial intelligence (AI) offers great potential here: modern speech recognition systems and large language models (LLMs) can efficiently transcribe, structure and improve the quality of reports.

However, the use of AI-based solutions raises key data protection issues – particularly with regard to professional
secrecy and the use of cloud services. In addition, it is difficult for many AI providers to estimate when systems of this kind will be classified as medical devices and which regulatory requirements they will need to meet as a result.

As part of the Innovation Sandbox for AI, the Office for Economy of the Canton of Zurich and the Center for
Information Technology, Society, and Law (ITSL) at the University of Zurich have worked with a number of experts to develop recommendations for the secure use of AI technologies in medical documentation.

The results are intended to help to reduce the administrative burden in the healthcare sector while taking into account the highest data protection and security standards.

The guide has been developed for providers of AI solutions in particular, but may also contain important information for hospitals, medical practices and other healthcare service providers.

View full article