AI Strategy for Healthcare Administration: Structure and Guidance in a Regulated Environment

Challenge
Navigating in the fog – a difficult start in the world of AI

In the public sector and the healthcare sector in particular, the use of artificial intelligence presents a unique set of challenges: on the one hand, there is enormous potential – for example, for more efficient administrative processes, better care and easier access to services. On the other hand, it involves particularly sensitive data, stringent regulatory requirements and decisions that directly affect people. The need to use AI responsibly and securely is therefore particularly high in this context. This makes the question of how organisations can navigate their first steps without getting stuck between the pressure to innovate and the need for caution all the more crucial.

For many organisations in the public sector, the first step towards artificial intelligence is therefore a huge hurdle; it often begins with regulatory issues. Often, the initial response to staff requests to use AI tools is a blanket ban, to ensure that strict data protection and compliance rules are not breached. Whilst this may appear to be a necessity, it does harbour a risk: bans frequently lead to what is known as ‘shadow AI’, where tools are used unmonitored and unseen in the background.

Furthermore, a classic chicken-and-egg problem often arises. Whilst steering committees are set up – which will play an important role in the use of AI – these committees do not yet have any specific use cases to oversee before the tools are introduced. At the same time, there is often a lack of willingness to invest until the potential has been demonstrated through initial pilot projects. The situation is complicated, on the one hand, by the fact that cultural changes in healthcare administration often take a very long time. On the other hand, it is compounded by budget cuts and, at times, a lack of prerequisites in the areas of technology, process clarity and data quality within a panel of doctors.

Approach to a solution
A holistic target operating model for measurable success

To address this complexity, we have not only developed a theoretical strategy, but are also working with KV to put the successful elements of our Target Operating Model into practice. This serves as a kind of ‘AI operating system’ for the entire organisation and answers the question of how AI will be embedded in the future. It sets out the path from overarching governance and management, through the selection and prioritisation of use cases, to operational implementation in day-to-day operations. A key part of this involved reviewing and further developing the cloud infrastructures to ensure the secure development and stable operation of the AI applications.

Implementation is based on three key levers:

Results and benefits
From vision to concrete roadmap

Thanks to this collaboration, the initial lack of direction has given way to a clear roadmap. The organisation now has a solid foundation on which to view AI not merely as a risk, but as a tool for improving efficiency.

• Clarity and security: The established AI governance framework minimises the risk of shadow AI and ensures compliance.
• Prioritised roadmap: There is a specific list of use cases, sorted by benefit and feasibility, which enables rapid implementation.
• Potential for efficiency gains identified: Particularly in the administration’s large-scale processes, areas have been identified where significant time and cost savings can be achieved.
• Organisational capacity: Through the communities and training programmes, knowledge is disseminated widely, enabling the organisation to grow independently alongside the technology.