AI and the future of health care

Symposium of the Society for Law and Politics in Health Care in February 2019

An event sponsored by the PTA Group, KENDAXA and Mbits

Is artificial intelligence revolutionising our healthcare system? This question was the focus of this year’s symposium of the Society for Law and Politics in Health Care (GRPG).

Although AI in healthcare is still a young field of research, there is already a wide variety of applications with enormous potential in terms of patient care. Once the ethical and legal issues have been clarified, there is little objection to their controlled use as a supporting tool. The biggest challenge, however, is to assess and ensure the quality of the data basis and the transparency of automated decision making.

The high-profile presentations met with a discussion-friendly audience. Already in the dinner speech on the evening before, Prof. Dr. Jonas Schreyögg captivated the audience with his remarks on the immense challenges health policy will have to face in the coming years. The demographic aging in connection with the (presumably increasing) shortage of skilled workers will further aggravate the already difficult care situation for patients in many places. The health care system in Germany is in an above-average position compared to other countries; no other system offers patients comparably easy access to medical care. One problem, however, is the relatively long length of stay in hospital after an operation, which results in a demand for specialist staff that can hardly be met. There is also a need for optimisation outside the hospital, especially in the cooperation between the various medical service providers. To improve the precarious medical care situation in rural areas, Schreyögg presented the idea of the “village nurse”. Equipped with extended competences in comparison to the current situation, the nurse could ensure the initial care of patients in close proximity to their place of residence.

On the symposium day itself, the applications of artificial intelligence in the health care sector were then examined in detail. Reinhard Karger from DFKI made it clear that all current applications of artificial intelligence are so-called “weak AI”, i.e. systems that support processes and methods and are particularly convincing in pattern recognition based on large amounts of data. “Strong AI” – these would be systems that “understand” interrelationships or develop approaches of “consciousness” – are still science fiction. According to Karger, AI can only ever support people constructively, but cannot creatively replace them.

But even with the applications of weak AI we are still at the beginning. There is no doubt that they hold great potential in terms of physical or cognitive assistance functions. These include diagnostic support for doctors, streamlining of processes in companies, relief from monotonous work routines and much more. However, AI has one crucial weakness: people can handle complicated issues by evaluating large amounts of data, but they are extremely successful at reducing complexity in advance – through their common sense. An AI is naturally not capable of doing this.

However, artificial intelligence is already being used successfully in many areas of healthcare, as Prof. Dr. Arno Elmer explained very clearly. In addition to the generally known applications such as chess playing robots and self-propelled cars, exciting applications are already running in routine operation in the healthcare sector. For example, doctors are supported by recommendations for action in cancer therapies and in the evaluation of image data. However, despite all the opportunities for the use of AI in medicine, it is extremely important to have good control over the parameters that serve the systems as a basis for suggestions: The right data must be available in sufficient quantity and quality, otherwise fatal errors can occur. However, the potential of AI for medicine is so great that it is definitely worth taking up this challenge.

In principle, the health insurance companies also value AI for supporting doctors and patients, especially when it comes to accelerating business and decision-making processes, reported Martin Litsch from the AOK Bundesverband. However, there are still a lot of open legal questions in connection with the transparency and traceability of AI-based decisions, which would have to be examined by experts. Ultimately, Litsch also considers an ethical debate to be indispensable. Data protection issues must be taken into account as well as the interests of patients. For example, there would be a need to discuss how the right not to know one’s own risks of illness can be guaranteed.

Indeed, procedures for the ethical evaluation of the use of AI in medicine are under development – since the 1990s, when the first medical expert systems were used. The progress made in the field of self-learning AI systems is bringing a new dynamic to the discussion. Georg Marckmann from the LMU Munich presented a procedure for the systematic ethical evaluation of AI applications which could serve this purpose well. It is important that each AI application must be assessed individually. An ethical recommendation can only be made if the actual need for decision support is proven and the algorithms and their data basis are sufficiently transparent.

Finally, representatives from industry also had their say: Ms Anna Bauer-Mehren from Roche reported on promising applications of AI in personalised medicine. In clinical trials, AI allows real patient data to be included in reference groups. This makes studies safer and speeds up approval procedures due to the better data basis. Here too, the quality of the data is crucial; they must be easily available, interoperable and reusable.

Mr. Michael Meyer reported on a unique initiative by Siemens Healthineers: With the help of an interoperability platform, measurement data from over 4000 institutions worldwide are already being brought together. The aim is to optimise services such as the maintenance of highly sensitive equipment and to make measurement data available to a larger community, which in turn will better support doctors in their diagnoses – these will be safer and can be better tailored to the individual patient. Siemens is also putting a lot of energy into the development of “digital twins” of patients with which treatments could be individually adapted and tested before use.

The Gesellschaft für Recht und Politik im Gesundheitswesen sees itself as a neutral platform for the scientific discussion of explosive health policy issues. For 25 years, it has been committed to the representation of health law issues in politics and contributes to the formation of opinion in the health care system by bringing together experts in an interdisciplinary way.