Increasingly complex treatment options and a growing number of patients are pushing doctors to their limits. Fortunately, the medical field has developed enormously in recent years. What used to sound like science fiction is now reality: AI uses algorithms to evaluate X-ray images, make diagnoses and suggest therapies. But how reliable are these results today? And what do doctors have to say? We’ll shine a light on the complex developments associated with the digital revolution in the medical sector.
Image analysis with AI
Cancer is still the second most common cause of death in Germany. Every year, around half a million people in Germany are diagnosed with cancer. One of the most promising applications of AI in oncology (cancer research) is the evaluation of imaging procedures such as X-rays or MRI scans. For example, AI systems can automatically identify the smallest nodules in the lungs and show them to doctors in a matter of seconds. AI is also already being used in prostate cancer diagnostics. An algorithm developed at the German Cancer Research Center (DKFZ) analyzes MRI images and could help to avoid further biopsies in the future. This is still a topic that requires further studies, but the initial results are encouraging.
Stronger together
A new international study led by the Max Planck Institute for Human Development in Berlin is particularly revealing. It compared more than 40,000 diagnoses from over 2,100 realistic clinical case vignettes. The result: the combination of human expertise and machine intelligence leads to the most precise diagnoses. “It works best together,” concluded the researchers.
Although the expertise of doctors is superior to machines in some areas, as soon as several AI models were combined, humans hardly stood a chance. They performed better than 85 percent of human individuals. However, hybrid teams of humans and AI systems achieved the best results. Stefan Herzog from the Max Planck Institute emphasizes:
“It’s not about replacing humans with machines. Rather, we should see artificial intelligence as a complementary tool that unfolds its full potential in collective decision-making.”
Cancer screening with artificial intelligence
AI can provide valuable services not only in diagnosis of cancer, but also in early detection and targeted screening. The results of mammography screening are particularly impressive: a German study showed that 18% more tumors were detected with the help of AI – without a higher rate of false alarms or unnecessary additional examinations.
The selection of those invited for screening could also be improved. Up to now, age has been the main factor in deciding who is invited for a mammogram. In future, AI could also take weight, tissue density or family history into account. Moritz Gerstung from the DKFZ explains:
“Ultimately, it’s a case of finding the needle in the haystack when it comes to early detection. […] And AI can help to pre-sort this haystack a little.”
Advise, not replace
In addition to image diagnostics, language models such as ChatGPT are also becoming increasingly important. According to Heidelberg dermatologist and AI researcher Titus Brinker, one in five doctors has already reviewed a treatment decision with the help of ChatGPT. At the same time, Brinker warns: “You should never trust them blindly, but always ask the AI about its sources and then check them yourself.”
A concrete example of a transparent system is the “Urobot”, a chatbot for urological questions developed at the DKFZ. It is familiar with all current guidelines and answered almost 90 percent of specialist questions correctly in a study – significantly better than many prospective specialists. Every statement is comprehensible, as the chatbot refers to specific passages in the guidelines.
AI in patient communication
AI could also make everyday life in doctors’ surgeries and clinics easier beyond specialist advice. Software for the automated creation of doctor’s letters already exists. Initial tests show that even patient questions can be pre-formulated by email using AI. However, for the time being the final control remains with humans.
In the long term, AI-based telephone assistants are conceivable – even with a synthetic voice. However, before such applications can be used in everyday clinical practice, they must be approved as medical devices. However, there is still a long way to go, as clinical studies take two to five years on average.
Practices that rely on the support of all-round solutions such as Doctolib can already use such systems. This provides doctors with a digital reception desk, including an AI-based telephone assistant that answers the phone and automatically recognizes patients.
But AI is also used in more acute situations. As the example of the soulfi app shows, AI can also be used as a first responder for everyday mental and psychological problems such as panic attacks, personal or professional stress, lovesickness or exam related anxiety. According to the developers, AI therapists are continuously trained by specialized psychotherapists on specific use cases in order to provide the best possible support.
AI in everyday hospital and practice life
Artificial intelligence is no longer just a topic in cancer medicine. As a joint study by the digital association Bitkom and the Hartmannbund shows, AI is already in use in almost one in seven practices. 15% of doctors in practices or medical care centers state that they use at least one AI-based solution – whether for diagnosis (12%) or practice management (8%). In hospitals, the use of AI has even doubled since 2022: 18% of hospital doctors use AI, especially for image analysis.
78% of the doctors surveyed see artificial intelligence as a great opportunity for medicine. At the same time, 76% are in favor of strict regulation. There is a high level of expectations – but also a need for security.
Other digital applications are also finding their way into everyday practice: 25 percent offer video consultations, 37 percent rely on online appointments. Robotics (26%) and virtual reality (11%) are also widespread in clinics, as are telemedicine and digital patient records.
Revolution with caution
AI has the potential to profoundly change and support medicine. It can make diagnoses more precise, improve early detection, make care fairer and reduce the workload of medical staff. Medicine is increasingly becoming a field in which humans and machines work hand in hand. Not only in cancer research. Even beyond oncology, digitalization in the healthcare sector is progressing with a high level of approval from the medical profession. But not without structural hurdles. For the change to succeed, not only technology is needed, but also trust, education and political decisions.
If this succeeds, the next major advances could be in preventive, personalized medicine. At the Helmholtz Center in Heidelberg, for example, research is being conducted into using AI to identify individual disease risks from clinical parameters and lifestyle factors – even before the first symptoms appear. This could mean the end of purely reactive medicine and the beginning of an intelligent, learning healthcare system that supports and treats people even before they know they need help.
cover image © Twopictures
Sources
Bitkom & Hartmannbund. (2025, Mai). KI in Praxis und Kliniken im Einsatz. https://www.bitkom.org/Presse/Presseinformation/KI-in-Praxis-und-Kliniken-im-Einsatz#_
Herzog, S., et al. (2025). Human-AI collaboration in medical diagnosis improves accuracy. Proceedings of the National Academy of Sciences. https://www.pnas.org/doi/10.1073/pnas.2426153122
MDR Wissen. (2025, Juni). Mensch und Maschine stellen gemeinsam die genauesten Diagnosen.
https://www.mdr.de/wissen/medizin-gesundheit/diagnostik-zusammenspiel-ki-und-mensch-am-effizientesten-100.html
Nature Medicine. (2024). Artificial intelligence in oncology: opportunities and challenges. https://www.nature.com/articles/s41591-024-03408-6
Statistisches Bundesamt (Destatis). (2024). Todesursachen in Deutschland. https://www.destatis.de/DE/Themen/Gesellschaft-Umwelt/Gesundheit/Todesursachen/_inhalt.html
Till, U., & Burkhart, E. (2025, Juni). KI revolutioniert die Krebsmedizin. SWR Wissen.
https://www.swr.de/leben/gesundheit/wie-ki-die-krebsmedizin-revolutionieren-koennte-100.html