Artificial intelligence and digital medicine: a view to the future

Professor Meder, you are currently in Stanford, where you have a visiting professorship. As a scientist in Germany (Europe) who wants to use digitalisation to advance your own discipline, do you have to travel to Silicon Valley to learn about the technological trends?

Benjamin Meder: Silicon Valley does not have the global reputation of an innovation region for nothing. There is an almost boundless optimism when it comes to the use of technologies. In particular, the cooperation between academic research and the private sector, as well as courageous startups, can serve as a model for Germany. Anyone who tries something new here, and even fails from time to time, is not looked at askance, but is valued as part of this extraordinary community.

What do you teach the students here, what can you convey from your work at the hospital in Heidelberg?

There is a lot of admiration here on the West Coast when it comes to our healthcare system and the social standards in Germany. But the quality and efficiency of our clinical sciences are also highly valued in California. “Made in Germany” stands for perfectionism in a positive sense. And German engineering – which includes medical innovations – is therefore highly trusted. So, there are some things that people like to report on.

Germany has a reputation for being very slow when it comes to digitalisation. How would you rate that in the field of medicine?

Medicine is unfortunately not a positive example here. The medical profession in Schleswig-Holstein has just found that the unprotected transmission of e-prescriptions via SMS or e-mail are not compatible with data protection. Such things could have been noticed years ago by the Society for Telematics Applications of the Health Card – Gematik for short. But now another stumbling block has been found in what is in itself a trivial approach to digitalisation. So, we’re practically going around in circles, and every innovation is being thwarted by excuses about data protection, costs or a lack of personnel. But even here in Palo Alto, digitalisation sometimes takes convoluted paths when simple, everyday processes are replaced by complex digitization nightmares. More courage is finally needed!

What are the biggest challenges? You complain that data protection hinders progress. Are there any solutions to this?

First of all, I would like to clarify that data protection is fundamentally not a problem construction, but it is an elementary right of every individual. So, in all our efforts, we must always consider the rights of individuals to privacy and data protection. Often, however, stakeholders in the health care system lack clear knowledge about data protection; regulations are sometimes imprecise and case law hardly exists. This creates a vacuum, an uncertainty that is readily used as an excuse for mismanagement and incompetence.

There are modern approaches to retaining better control over data: by bringing algorithms to the data, for example, rather than the other way around. After all, the money is counted in the bank and not first collected in a large external hall and added up there. In Heidelberg, Prof. Engelhardt is working very intensively on the topic of federated learning.

Not only are you in charge of a top hospital in Germany, but you are also involved in many projects involving the use of informatics, of artificial intelligence in medicine, especially in interventional cardiology. Where does this enthusiasm for informatics come from? What progress do you hope to make?

I already experienced incredible possibilities of digitalisation in school and later during my civilian service by programming my own projects. I personally believe that the invention of the wheel, writing, or penicillin are on par with the possibilities of artificial intelligence and digital medicine. However, since it is such a broad, sometimes unmanageable, as well as dynamic system, I believe that it needs people who bring order and overview into it. That’s what I enjoy and that’s what excites my young team members. Presentations by AI experts usually have a much larger audience than on other topics.

In industry, including at LAPP, people are working with the “digital twin” to improve a real-world product or process. Interventional cardiology is also about the “virtual image” – are there already results?

We co-founded the digital twin, so to speak – during earlier work on complex heart simulations, together with a medical technology company. Today we have projects on the heart twin directly in patient care, e.g. in electrophysiology or complex heart defects. You can learn a lot from a digital image of reality, which details really contribute and then influence the whole thing in a targeted way or optimize it. So, in medicine today, we already use complex reconstruction and simulation for the success of a subsequent intervention.

Cardiovascular diseases are among the most frequent causes of death. What can patients hope for as digital cardiology advances?

I think AI systems will primarily ensure that we don’t lose sight of the multitude of information on each individual patient. In addition, the algorithms will have to integrate important and less important information into decision-making processes in order to maintain a level of quality across the board – today, this is only possible with a great deal of personnel effort. But there is also a lot of untapped potential in prevention, which we can hopefully leverage in the future through clever apps and digital communication networks. If we manage to make social behavior in our population a bit healthier through digitalisation, this will have a strong impact on cardiovascular mortality.

Does every physician today also have to be a computer scientist – does this perhaps even apply to every subject: computer science as a “must” training?

It would be presumptuous to call myself a computer scientist. My colleagues in computer science have completed a full course of study and therefore have a completely different level. But it never hurt to think outside the box and to have some command of the language of other disciplines. That’s exactly what we try to teach in eCardiology. After all, we teach physics and chemistry to some extent in medical school – so why shouldn’t we teach anything about programming machine learning algorithms?

When it comes to AI, utopia and dystopia are often close together. When it comes to medicine and health, tempers flare especially quickly. What ethical regulations do you think are necessary to ensure that people retain their trust in our medicine and doctors?

I think it’s the values that we have to convey so that changes and new things are used in the best interests of society. If we as physicians and society do not manage to place these fundamental values above innovations, a utopia can also become a dystopia. Therefore, we need a constant discourse about what we want versus what is possible. Unfortunately, I think we will see many more examples where we will have really negative experiences with artificial intelligence. We should anticipate these scenarios and keep their frequency as low as possible. Overall, however, I firmly believe that the benefits of artificial intelligence for medicine outweigh the disadvantages and that the increasing scarcity of resources cannot be addressed in any other way.

On Artificial Intelligence, there is a lot of imagination being awakened and encouraged in the film industry. Do you yourself love science fiction and if so, which is your favorite movie and why?

I like science fiction movies. “2001 Space Odyssey” by Stanley Kubrick is a true masterpiece. As early as 1960, he excellently foresaw many aspects of artificial intelligence and possible challenges, or rather undesirable developments, and implemented them cinematically in a brilliant manner. The artificial intelligence HAL 9000 develops a consciousness and a certain momentum. It successively opposes the interests and values of the human crew. Overall, I think this is a possible scenario on the dystopian side; in the film, the AI has created its own values and goals and competes with the “protein computers” – that is, us humans.

To conclude: If you were allowed to write (or commission) a screenplay – what story, what plot should the film tell? (Where should it be set)

The film is definitely set in the hospital. I come to rounds, take my iPad and have visualized data and results from each patient – at the touch of a finger. I can easily request prescriptions and exams – just like an online order. Results from external findings and reports are seamlessly integrated and clearly displayed. But that’s just science fiction.

Thank you very much for this interview!