Interview | Prof. Dr. Alexander Alscher – Founder and Director of samedi and Holder of the Professorship for International Management at BSP Berlin and Dr. Christian Herles – Lawyer and General Counsel at samedi

1. When did you start dealing with the topic of artificial intelligence at samedi and what role does AI play for samedi's software today?

Prof. Alscher: This is an ongoing development and ultimately also a question of how AI is defined. So far, such a definition does not exist yet. Even the EU AI Act does not conclusively and comprehensively define AI. In this respect, we at samedi have been dealing with things like this right from the start – even if it was initially more like smart algorithms that I wouldn't describe as AI – for example in the configuration of our resource planning, where our software automatically checked for inconsistencies and alerted the users to them. About six years ago, we hired our first data scientist, who helped us to more intensively deal with our data. In the broadest sense, this was the beginning. Since then, we have systematically expanded this branch. In the meantime, our previous data team has turned into our AI team. To expand our offerings, we use self-hosted AI (LLMs) to comply with data protection rules. In our work, we are currently going through three phases: Firstly, we implement AI within our company so that, secondly, we can offer good and complete solutions to our customers. The third step is to provide patients with more efficient applications.

2. What benefits do AI-supported systems in central applications have for patient sovereignty? And are such systems necessary?

Prof. Alscher: In the context of increasing digitalization and the associated prevalence of data, it is important to have such supporting AI systems. AI in particular can help us to better shape patient sovereignty. One example is the patient account we have at samedi, where patients can decide for themselves which data they want to share with which doctor and how they want to be notified. However, we also believe that at a certain point it becomes difficult for patients to keep track of everything and understand it down to the last detail. At this point, AI can simplify management of your own data through recommendations, for example. In addition, chatbots can quickly clarify comprehension questions. And many other things are conceivable, so such systems can be highly beneficial.

3. Providers from the USA currently dominate AI applications in the healthcare sector. What impact does this have for the data of European patients?

Dr. Herles: In Germany and Europe, we are strongly focused on the application side, i.e. on innovations in the area of healthcare apps, for example, or in the provision of care by service providers. A significant weakness is that we lack large data centers and storage locations in Europe and Germany – and therefore also large data models. The USA with its tech giants is clearly at an advantage here, and the European market depends on the technological infrastructure existing over there. The data of European patients is protected by the General Data Protection Regulation. This also means, however, that if I transfer data to the USA because I use a US-based processor, this is subject to strict conditions and more difficult to implement than in the EU's internal data market. The further development and implementation of AI on the European market could therefore be even better if we had a comparable infrastructure over here.

4. The AI Act enacted by the European Union classifies AI applications into categories ranging from minimal to unacceptable risk. Although health and medical applications were already regulated, they now fall into the high-risk category of the AI Act and are subject to further requirements. What challenges does this create for the development and implementation of AI systems in the healthcare sector?

Dr. Herles: Contrary to the entire medical treatment context, AI has so far not been regulated on a technology-specific basis. The EU AI Act has now at least partially defined AI. This means that a statement is made, for example, about what distinguishes AI from a computer program that uses normal algorithms. Legislators have now adjusted to the fact that although humans define the purposes of AI and determine which data is used, AI produces results that influence humans in their further actions. The legislation is therefore based on black box logic, which is a big step legally, because it recognizes the risk of AI. This risk is now to be minimized through various requirements. On the practical side, this also results in challenges such as bureaucracy. In addition, at a national level, there is a lack of concrete implementation and corresponding guidelines for dealing with AI. To find such guidelines will be another challenge. Ultimately, it will also depend on how the AI Act is implemented by the individual Member States.

5. What are the potential risks and opportunities arising from the new regulation for patients, healthcare providers and the industry?

Dr. Herles: For me, three aspects are central here: Firstly, the EU AI Act creates legal security. The regulation also provides for a lot of conformity assessments, certificates and the like, which also create a basis for trust in AI applications. A second aspect is the fundamental improvement in the quality of AI. This is because the risk assessment also results in measures for increasing security, such as regular training, monitoring and much more. The third aspect is transparency: Providers, manufacturers and users have to disclose where and how they use AI. Providers have to explain how their AI works and what data it uses. This enables patients to understand how results are achieved.

Prof. Alscher: KI has enormous potential to optimize existing processes and generate significant progress, notably in the medical sector. One way to let everybody profit from AI in the future is to speedily address risks already known to us today – such as false data bases, prejudices or hallucinating, i.e. the production of fictitious results – and to establish processes and quality checks through regulation, thus designing a more standardized way of working on and with AI.

About samedi:

samedi – We organize Health. Since 2008, Berlin-based e-health company samedi has been digitalizing clinics, medical care centers and practices as well as other service providers in the healthcare sector. As flexible Software-as-a-Service (SaaS), the company’s e-health and patient portal solutions (samedi healthspace) enable efficient internal and cross-sectoral digital patient management. The company's mission is to make treatment coordination with online services simple and secure for everyone involved. All medical organizational processes – including appointment management, patient admission, integrated resource planning, treatment management and accompanying patient communication – are mapped and managed by samedi’s solutions with the aim of automation, error reduction and quality improvement. Maximum data protection and a wide range of interfaces ensure seamless integration into existing processes and systems. Regardless of institutional size and specialization, processes are digitized smartly and holistically with tailor-made solutions in more than 6,000 healthcare facilities for 35 million patients in the DACH region.

Further information:

• Website samedi: www.samedi.com/en/
• The European Parliament on the AI Act: www.europarl.europa.eu/topics/en/article/20230601STO93804/eu-ai-act-first-regulation-on-artificial-intelligence

You are interested in AI and the healthcare industry? Cluster HealthCapital – Gesundheitswirtschaft Berlin-Brandenburg supports institutions in the capital region free of charge on many topics relating to financing, funding, innovation management, cooperation partners, skilled professionals, real estate and sustainability. Through our Newsletter and LinkedIn channel we provide you with helpful information. In addition, we offer a newsletter dealing exclusively with digital health issues.

You are interested in/are developing AI or robotics for the healthcare sector? We are an associated partner in the TEF Health project and distribute relevant information through our network.

Feel free to get in touch with us.

healthcapital@noSpamberlin-partner.de
Tel.: 030 46302 – 463