back

»Digitization without cyber security is unthinkable – especially when dealing with data as vulnerable as that collected in healthcare.«

Prof. Dr. Ulrike Attenberger is Director of the Clinic for Diagnostic and Interventional Radiology at Bonn University Hospital (UKB) and head of the KI.NRW flagship project »Innovative Secure Medical Campus UKB«. We asked her about the topic of AI in medicine.

What is the aim of the »Innovative Secure Medical Campus UKB«?
The aim of the ISMC is to rethink patient care holistically using the example of a university-based maximum care provider. In view of the demographic change and the resulting shortage of specialists, as well as the increasing cost pressure in the healthcare system, there cannot and will not be a »business as usual«. In 15 subprojects, the ISMC demonstrates how digitization and AI will take patient care to a completely new level in the future, optimizing the use of available resources, reducing the workload of medical staff, and improving the precision of diagnosis and therapy.

What tasks can AI systems already perform in the medical sector?
AI systems are of course a very broad term and you have to define very precisely what you mean by it. In the field of radiology, for example, AI is already being used in various places, such as in the acquisition of MRI images, in order to further increase quality. AI systems can also already partially analyze image data and alert physicians to potential abnormalities. However, AI does not replace the attending physician, but rather serves to support the diagnostic process. As of today, the networked and integrating use of AI systems beyond fields of application in individual specialties has not yet been achieved with regard to the healthcare system as a whole.

What are the fields of application in medicine in which AI systems will be increasingly used in the next few years? Where do you see the most potential?In my opinion, the fields of application within workflow optimization can be divided into three areas: i) the intelligent control of patient flows to reduce waiting times as well as the deployment planning of personnel and (large) equipment, ii) the optimization of diagnostics, and iii) the final therapeutic path in the sense of individualized precision medicine. In addition to the use of AI in diagnosis, the vision of a digital twin of patients, for example, holds great potential in this context. Here, therapies could be pre-tested for their effectiveness on the digital twin before they are used on humans. However, this is certainly a vision whose implementation is still a distant reality. The implementation of extended reality environments is much closer to reality. These are used for optimized planning and execution of operations or also for improved communication with our patients by helping the attending physicians to communicate complex clinical pictures or therapy approaches more clearly than is possible today.

Many people wonder: as a doctor, can you trust an AI blindly? How is it ensured that no wrong decisions are made here?Ultimately, the final decision will always be in the hands of the appropriately qualified physicians for the foreseeable future, at least according to my assessment as of today. In addition, intensive work is currently being done to develop possibilities that make the decision-making of an AI algorithm comprehensible – a keyword in this context is »trustworthy AI«.

The use of AI systems requires a lot of data – in your case, sensitive and personal health data. To provide special protection for this data, you work closely with the Bonn Cyber Security Cluster. What measures are taken to protect the data?

Digitization without cyber security is unthinkable – especially when dealing with such vulnerable data as that collected in the healthcare sector. In this lighthouse project, we are working with strong partners according to the »security by design approach«, in which state-of-the-art AI technologies are combined with the issue of cyber security and data protection from the very beginning. Sensitive and personal data must only be accessible to a very limited group of people. To this end, this data is stored in highly secure environments and protected from potential threats by monitoring systems.

How do you think patients will respond to the use of AI?
Communication plays an important role here. It must be clear to patients what contribution AI makes to diagnosis and therapy and how medical staff can control AI systems. In the end, however, it is the success of the treatment that counts – whether with or without AI will probably play a secondary role for our patients.

With a view to the shortage of skilled workers in your industry: will the use of artificial intelligence be of any help here?
Artificial intelligence will play a decisive role with regard to the shortage of skilled workers. For example, in the creation and management of work schedules and appointments for our patients. This will inevitably also lead to a reduction in the workload of, for example, nursing staff or medical-technical radiology assistants. In the future, new robot technologies will also provide additional relief in the nursing sector and in large-scale equipment medicine, e.g. in the transport or bedding of patients. In summary, the meaningfulness of the use of artificial intelligence must always be measured in terms of the time saved for the treating staff; time that is then not lost in the organization, but is available for patient care – from one person to another.

What do you wish for the medicine of the future?
In order to shape the future of medicine in the best possible way, I would like to see an even more efficient use of resources. My vision is that target-oriented, AI-supported diagnosis and therapy paths produce a maximally optimized therapy result for our patients without loss of information at the interfaces of the various treating units. The »organization time« for the staff should be reduced in order to have more time for the patients*. In addition, improved communication between the various disciplines and in doctor-nurse-patient contact is desirable.