Professor Roland Eils is an acknowledged expert for bio-medical computer sciences, genomics and personalised medicine. He was head of the department ”Theoretische Bioinformatik“ [Theoretical Bio-Computer Sciences] at the Deutsches Krebsforschungszentrum (DKFZ) and in the area of “Bioinformatik und Funktionelle Genomik“ [Bio-Computer Sciences and Functional Genomics] at the Universität Heidelberg, prior to taking over as Head of “Digitale Gesundheit“ [Digital Health] at the Berliner Institut für Gesundheitsforschung (BIH) in April 2018. The researcher is aiming for ambitious goals: Eils intends to develop the research area in Berlin to become a mutual data space and to scale the infrastructure of digital health such that the huge amount of data coming from research results and medical care data can be used efficiently.
Professor Eils, Digital Medicine is considered to be the medicine of the future and is to facilitate - amongst others - doctors and patients to achieve more reliable and more precise diagnoses. How and with what technologies are you going to tackle this subject?
The section of digital medicine which we want to get involved in especially in Berlin concerns two data worlds that exist parallel to each other and which we want to link to each other: on the one hand data from the clinic where they are raised mainly while caring for patients, and data from basic bio-medical research on the other hand. There is a disconnection at this point which we want to bridge over in order to link these two data sections. In doing this we try to develop methods and technologies such that this data flow becomes transparent and easy to understand for the attending physician as well as the researcher. But we also deal with questions about ethic and regulatory marginal conditions and want to address in this way possible objections on establishing such data flow. The added value we intend to achieve here would work in both directions: On the one hand we would contribute to improve the patients’ care due to research. And on the other hand we could benefit on the basic research side by observing how our results are implemented into the care and what results will be achieved concerning the patient. In this way we will we gain a better understanding in how far our illness models used in our basic research work at all in caring for a patient.
You have recently responded to the call from Berlin. How do you assess the location as a centre for technological innovation in particular in respect of Artificial Intelligence? Where does Berlin differ compared to other locations?
There are several reasons why Berlin is suitable as a model region for Digital Health. The inpatient care in Berlin is publicly owned with a share of above 40 per cent. The Charité is not only an internationally renowned university clinic, but by introducing the Charité-owned Health-Data-Platform an important basis was laid for digitalising and linking all care-relevant data. At the same time the state of Berlin owns Vivantes, the largest communal clinic group in Germany.
With the Big Data Centre and the Deutsche Forschungszentrum für Künstliche Intelligenz [German Research Centre for Artificial Intelligence] as well as a highly dynamic start-up scene, it appears that the IT-competence in the health sector has accumulated in Berlin. Nearly every third German AI-enterprise has settled in the capital.
How significant is the BIH and its research for the AI industry in Berlin and beyond?
Currently the BIH [Berlin Institute of Health] is most definitely one of the most important players in the sector of Digital Health. My professorship for Digital Health is one of the first of its kind in Germany. Robert Gütig holds a BIH-professorship for Artificial Intelligence, Silvia Thun a professorship for Data Interoperability. The Digital Accelerator means that the BIH has a highly professional infrastructure available to support young company founders within the Digital Medicine to set up their own start-up company. Subsequently this means that the Berlin Institute of Health has not only recognised the signs of the time but has also consistently implemented them.
What is the situation like in respect of an infrastructure and cooperation on the subject of Digital Health in Germany?
Many processes within the health care are still processed on paper or are compiled in data silos in institutions that do not communicate with each other. Viewed technically, all this could be wonderfully linked as there is no such challenge in the sense of ‘this cannot be solved and be technically implemented‘. The restrictions have grown historically which means that there are IT-solutions available in the various departments of a clinic; there are different IT-systems well beyond the restrictions of a clinic. There exist many aspects that will have to be addressed and overcome in order to ensure that these data worlds between different data silos and various institutions can even talk with each other. The exchange of data as such is first of all a mere technical exercise of fingers. However, the description of a specific illness must be carried out in a clinic up North the same way as in a university clinic in other parts of Germany to make the data between patients and institutions comparable and exchangeable. And the challenges are definitely to be found in the section of the so-called interoperability of data that are particularly challenging in the health system.
Apart from the issues of your own field of research what do you think will be the challenges, problems and potentials for the future of Deep Learning, Machine Learning and Artificial Intelligence in general?
In respect of the potentials for the future the only answer can be: Everything is possible! The challenges currently certainly include the question - discussed manifold - to what extent machines that we are teaching the so-called Artificial Intelligence with the assistance of programmes, can make independent decisions that cannot be controlled or retraced anymore by human beings. The Artificial Intelligence is a machine-based intelligence which will no doubt be able to develop further on its own. And it is my opinion these are real social challenges that we must face today: What sort of competence do we want to grant these machines with their artificial intelligence?
How can we guarantee that we remain master of these systems also in future with its progressing technology and IT-developments? And also, how can we implement control mechanisms in order to regulate what machines are allowed to do and must not do? The potentials are very large indeed. I am sure that with a progressing development, machines will be able to carry out tasks at least as well as human beings. But the question is: What happens then? What happens to all of us human beings? This is not a philosophical question but will become a question that really exists in ten or twenty years’ time. And my impression is that we do not really deal well with these questions today. We seem to have pushed this slightly towards the accompanying researches such as ethicists and philosophers. But we should really view this from a point of view of research as well.
What would your recommendations be for people and institutions that are dealing with Artificial Intelligence in order to promote innovations in their specific area?
We think it wonderful, of course, when people get interested in the opportunities of using AI in the health sector. We are very happy about every start-up that offers innovative ideas. And above all, we consider the Digital Health Accelerator of the BIH to be a wonderful platform to bring together, promote and support these interested groups and communities to take these developments forward ready to join the market.