Doctor to AI & Co.: “Digital tools must remain in the hands of medicine”

Contents

Artificial intelligence is supposed to predict diseases, health apps to accompany therapy, and large amounts of data to accelerate research. The promises of digital medicine are great, but reality looks different. Doctors are overworked, further training is possible under difficult conditions, and industry is putting pressure on. This also became clear at a press conference of the German Society for Internal Medicine (DGIM). Digitization also raises fundamental questions, for example, about the quality of evidence, access to data, and the future role of the doctor.

While some want to drive the digital revolution forward, others warn of a collapse of the human factor in the system. Staff shortages, an overwhelming hospital reform, and increasingly complex diseases are pushing clinics and practices to their limits – an environment in which digital tools are actually supposed to provide relief.

Doctors have little time

At the forefront of the crisis is the next generation of doctors. Dr. Lena Levien from the YOUNG DGIM working group describes a daily routine characterized by overload. The planned hospital reform with its increasing outpatient care is exacerbating the pressure, as the necessary infrastructure and personnel are lacking. The result is that doctors have hardly any time, especially for further training. “This add-on, meaning this teaching, [is] simply a luxury that not every department can afford anymore,” warned Levien. The consequence is frustration for both trainers and junior doctors. Manual skills like a gastroscopy cannot be learned from books. The lack of time for teaching and research endangers not only the future of doctors but also the quality of care.

The DGIM chairwoman, Prof. Dagmar Führer-Sakel, sees clinic directors as responsible for “creating free spaces” here. She emphasized that research is not an optional extra: “Science is not an option, but the prerequisite for us to be able to act innovatively, effectively, and responsibly tomorrow as well.” Scientific thinking is a basic requirement for making well-founded decisions in a medicine shaped by AI and data.

Hope for real-world data

According to Linea Schmidt from the Hasso Plattner Institute (HPI), the linking of different data sources is important here. Her project aims to connect billing data with clinical information from electronic patient records. Only this combination enables a comprehensive picture of the health trajectory. However, Schmidt emphasized that this process is anything but trivial: “Linking different data sources is often a bigger challenge in the healthcare system, not only in Germany but also in the USA, than one initially thinks.”

She sees great potential in the unstructured data of the electronic patient record (ePA), from which, for example, disease trajectories can be more precisely understood and risks identified earlier using Natural Language Processing. Together with the Mount Sinai Health System, she is researching a project with over 500,000 hypertension patients in the USA. The goal is to “find out which patient groups benefit particularly from remote monitoring and how health trajectories change.”

The HPI criticized “wellness apps” that make medical claims without approval. “And you definitely have to be careful and see that more emphasis is placed on taking action against this. What are you actually allowed to state in your app description if you are not a regulated medical product, so as not to mislead patients.”

Real-world data are important for generating evidence. The Federal Institute for Drugs and Medical Devices (BfArM) primarily relies on randomized controlled trials (RCTs) for the approval of Digital Health Applications (DiGA). “I believe we can still work on ensuring that this is not exclusively the gold standard, but that there are also situations where conclusions can already be drawn about whether such applications work or not based on real-world data from everyday care,” the researcher demanded.

Industry demands free data access

The pharmaceutical industry supports the demand for data. Dr. Claus Michelsen from the Association of Researching Pharmaceutical Companies (vfa) emphasized that access to data is a crucial location factor. He warned against making it more difficult through payment models: “In this respect, I don't think Germany would do itself a great favor if it started developing cost models for such data access.”

The BfArM has already received initial applications for access to research data from both industry and research. However, there are still difficulties with data preparation that is friendly to data protection.

Research purpose is decisive

The decisive factor for a positive decision by the BfArM and the release of data from the Health Research Data Center is solely the research purpose, not who is applying for access. However, ethicist Jessica Hessen cautioned in an interview with heise online that it is “hardly possible to distinguish anymore what is actually health data for the common good or what is only for commercial interests.” Insured people cannot decide whether they want to contribute to Google, OpenAI & Co. improving their products.

For the industry, Germany has fallen behind in global competition with the USA and China. There is a lack of venture capital and, above all, speed. “I would wish for speed in translation, meaning the courage to implement things quickly. I believe that is what can be learned most from other economic areas,” said Michelsen.

Who maintains control?

Despite all the euphoria for data and AI, Prof. Georg Ertl, Secretary General of the DGIM, pointed to the enormous but unstructured treasure trove of data that is also being created by the ePA. “We have an infinite amount of data,” said Ertl. However, he clearly opposed the industry's demand for free access: “Ultimately, it is very clear that access to this medical data will not be free and cannot be free either.” Collecting, structuring, and archiving is complex.

Above all, Ertl warned emphatically against a loss of control. “That all these tools must remain in the hands of medicine, I am an ardent advocate for that, if it falls into the hands of computer scientists or, even more so, industry. Then I see dangers there, not only in quality but also in the ethics of these applications.”

In the end, according to the consensus of the medical professionals, technology must not undermine the foundation of care. The doctor must have the final say, and patient trust must not be jeopardized. Ertl concluded with an appeal that highlights the limits of digitalization: “Yes, I believe it will be crucial that our population, our society, retains its trust in medicine. And that is based on the doctor-patient relationship, and that must be maintained and actually always further improved. […] The ethically and medically comprehensible decision made jointly by doctor and patient will, I hope, continue to shape our healthcare system in the future.”

(mack)