ai in medicine

Lucas M. Bachmann

Founder / CEO

medignition AG


Artificial intelligence (AI) is currently on everyone’s lips. However, it rarely seems to be truly practical in medicine. Our CEO Lucas M. Bachmann and his co-authors recently published a position paper on this topic in the prestigious Frontiers in Digital Health: Artificial Intelligence and Statistics: Just the Old Wine in new Wineskins? In a brief interview, he gives us his view of things.

Lucas, numerous innovations today rely on artificial intelligence. It seems to be a permanent fixture in all industries by now. You are a bit more skeptical about the use of artificial intelligence in medicine. Why is that?

Artificial intelligence and the associated sub-field of machine learning certainly have great potential - this is shown both by the interest of the general public and the number of scientific publications published on this topic. In clinical research, however, the terms artificial intelligence and machine learning often lead to confusion. There is both a lack of consensus regarding their use and a lack of a standard for how the results of clinical studies are presented. Studies are therefore often difficult to compare, if at all.

Where do you see limitations in the practical use of artificial intelligence in medicine?

Machines generally need huge amounts of data to be able to handle tasks independently and make predictions. The data provided, in turn, must be of high quality in order to avoid false statements. These requirements are not met in many areas of medicine. Electronic medical records, for example, have incomplete or even erroneous data, and clinical research often works with smaller data sets. In addition, health care is strongly influenced by political and economic factors as well as medical practice norms. It is therefore necessary to create a framework that enables the widespread use of artificial intelligence first.

Despite all the skepticism, you believe AI has great potential. Where do you locate it?

While the ability of humans to learn new things through experience is limited, AI algorithms are constantly learning through new data. Therefore, in clinical areas such as pathology or radiology, which have large and well-structured data, significant progress has been made using AI. I therefore see the benefit of AI for healthcare in, among other things, its use in pattern recognition - for example, in patient triage using image analysis. Efficient, automated triage could significantly reduce the burden on the healthcare system. In addition, AI applications can make sense in those places in the world where there is a great shortage of medical experts.

What needs to be done for AI to realize its potential in the healthcare industry?

First, the various stakeholders need to agree on the definition and use of AI terms. If the different areas speak the same language, findings can be consolidated and the requirements for research defined. The latter, in turn, should be designed in such a way that the results flow directly into practice. And, of course, the legal framework must be created to bring AI into widespread use. Great attention should be paid to the collection and maintenance of relevant data. The hope that good data for the development of AI applications will automatically accrue through clinical routine is an unrealistic dream.