HiGHmed is one of the four consortia of the Medical Informatics Initiative (MII) selected for funding by the German Federal Ministry of Education and Research (BMBF). It is coordinated by Prof. Dr. Roland Eils, who is the founding director of the Center for Digital Health at the Berlin Institute of Health (BIH) and the Charité. He also heads up the Health Data Science Unit at Heidelberg University Hospital. An expert for biomedical informatics, genomics and personalised medicine, Eils uses methods of artificial intelligence, big data analysis and mathematical modelling. In addition to his academic activities, he has established two companies, both of which are active in the field of digital health. We spoke with Professor Eils about the HiGHmed consortium’s approach.
Mr. Eils, what specific approach does HiGHmed pursue? How does this differ from MII’s other consortia?
Roland Eils: What is different about HiGHmed is the way in which we describe medical issues such as disease indications or images. We use the openEHR standard for this and are the only ones to work with this open standard to this extent within MII. We chose this standard because it is used in many places in the field of digital health. As a consequence, countless standardised, interoperable descriptions of medical facts already exist for us to use. What is more, we can export from openEHR to other standards, such as the FHIR standard used by other consortia. We also take a different approach by prioritising the roll-out phase, which is actually only due to begin in the next funding period from 2022. MII only links funded university hospitals. In contrast, we have already begun including private and regional hospital operators in HiGHmed now. This sets us apart within MII.
One of the HiGHmed use cases is oncology. Could you briefly describe the procedure and objectives? How far have you got here so far?
Eils: Within the oncology use case, we wish to expand virtually what has already been developed at the National Center for Tumour Diseases (NCT) in Heidelberg to the participating university hospitals. We have established so-called molecular tumour boards at these, which include standard diagnostic and diagnosis-relevant data (from pathology and radiology, for example) along with “research-based” data from high-throughput genome sequencing or modern imaging methods. These boards enable more precise treatment decisions. In HiGHmed, we are now also developing a platform for virtual molecular tumour boards. By extending the Heidelberg concept to a network of eight university hospitals in addition to private and regional hospitals, an even larger database is created. We are currently in the process of implementing the corresponding documents, scenarios and data types in openEHR. In the next step, oncologists and radiologists can fill the boards with data. We want to make a functioning platform available for use in routine care as early as next year.
You are also active in the International Cancer Genome Consortium (ICGC). What are your experiences here? What role does the quality of biomaterial samples play in high-throughput studies?
Eils: The quality of the samples is extremely important for everything you do with them. If the quality is poor, the data we collect from the samples is only of limited value. For us, it is very important that the pathological information on samples is accurate – that we receive precise information on which part of the tumour a sample originates from and how high the cell and stroma fractions are, for example. A fully quality-assured provision and annotation of samples is indispensable for our work.
To return to HiGHmed: how do you involve stakeholders in your developments?
Eils: As previously mentioned, we have already integrated regional and private hospital operators into HiGHmed. This sets us apart within MII. The patient stakeholder group is also very important to us. Various patient organisations and representatives are involved in our consortium. This cooperation is extremely important for us in order to obtain feedback on relevant questions – how we should inform patients or what additional tests we can justify performing on them to, for example. With regard to doctors as another stakeholder group, we are in close contact with both the state and national medical associations. We also wish to involve general medical practitioners in our activities and to find out what is of particular interest to them and where the pain threshold lies in terms of additional annotations. In June, we will host a symposium on electronic health records that will bring together representatives from patient organisations, the medical associations, and health insurance companies. We wish to know from the health insurance companies what their ideas and models are; this includes the models that they have already implemented for electronic health records.
How do you ensure data protection and security?
Eils: This is a major issue within all consortia and all take this equally seriously. On the one hand, we have developed a data protection concept that has been endorsed by the TMF. On the other, we began working on the ethical and legal aspects of the total sequencing of the human genome within the EURAT project in Heidelberg before the Medical Informatics Initiative was even launched. We have developed guidelines on how to treat such data as well as templates for informing affected patients. We are now benefiting greatly from this preliminary work.
What do you think of the cooperation between the Medical Informatics Initiative, German Biobank Node and German Biobank Alliance?
Eils: I think it is very important that the biobanking activities of GBN and GBA are linked to those of MII and naturally also HiGHmed. We should avoid treating the standardised description of samples differently at each site in all consortia and instead agree on a minimum data set and common standard. In my opinion, it is of little consequence whether openEHR, FHIR or another standard is selected as long as we ensure that our data is interoperable. Fortunately, representatives from GBN, GBA and MII are already collaborating on joint activities, especially in the modelling of oncology-relevant annotations. We’re on the right track here. However, I believe that this cooperation should be formalised to also establish a national precedent. So that what we develop jointly becomes a guideline for the whole of pathology in Germany.
The interview was conducted by Verena Huth.