Creating good standards

To bring medical data together for patients’ use: That is the goal of the association MIRACUM (Medical Informatics in Research and Care in University Medicine) in which the University Hospital Freiburg and the Department of Medicine at the University of Freiburg are participating.  With 4.4 million euros in funding, Freiburg is the second strongest of the eleven university hospitals nationwide. During an interview with Jürgen Schickinger, the private lecturer Dr. Martin Boeker, director of the working group medical informatics in the Department of Medicine, explains what progress MIRACUM hopes to make.

Source: everythingpossible/Fotolia

Herr Boeker, why does medical data have to be bundled at all?

Martin Boeker: In our consortium we have a total of more than eleven million patient entries. Most of them are not standardized and distributed amongst various locations. While a large part can be found in our computer center, such as data pertaining to diagnoses and treatments, raw data, medical data and all kinds of findings such as X-rays or CT scans can be found in multiple places. In addition, it includes data from clinical and basic research. A lot of these data sets have various formats and they often work with different terms. For instance, one time it is called “blood sugar” and somewhere else it is called “blood glucose”. But the same value is meant in both cases. We want to bring all the data together and standardize it in such a way that the formats, terms and measurement units coincide.

Can doctors or researchers already access the data?

They can in most cases, but not fast enough and not always in a targeted fashion. Freiburg does have a very good electronic patient filing system in its hospital IT system. But we want to adapt our data sets even more, making it more transparent. It has to be clear what you are asking content-wise – which of the more than 100 blood sugar levels, which tumor marker, which parameter? The answer should cover as many data sets as possible that include the search term.

How will patients benefit from this effort?

It is a step toward personalized medicine – to treat patients according to their individual needs. Targeted inquiries deliver all patients, for instance, with a very specific tumor that also have a very specific tumor marker. We can see to which type of treatment this subgroup responds best. We can then consistently improve therapies for them and develop new approaches. That wouldn’t only be progress for someone with cancer, but also patients with significant widespread diseases such as chronic obstructive pulmonary disease (COPD) would benefit from a more exact diagnosis and thus a better treatment. Further, researchers could recruit patients according to certain criteria for clinical studies. In the end, that helps patients too.

Medical informatics in conjunction with electronic patient files and personalized medicine is becoming more and more important, says Martin Boeker. Photo: private

Who will merge the data and standardize it? Computer programs or people?

Both. A lot of medical data is not even coded, such as information located in doctors’ letters and similar texts. For that we are working together with a company called Averbis, a spin-off company from the University of Freiburg. The firm has voice recognition technology that can extract such information from texts. This cooperation is even being developed into a national project. Lab findings, on the other hand, are well structured. They consist of values from tables and are basically easily summarized with programs. But before we do anything there, we have to standardize the terms and references as well as test methods and measurement units. People have to do that.

For which of these tasks is Freiburg specifically responsible?

We want to make the voice recognition technology available to all of our consortium partners. Furthermore, we are building a special interface for the hospital information system. It is meant to enable us to find every patient with a specific tumor, for instance, with a very specific pattern of markers. We also want to bring data from genetic analysis of tissue samples to tumor boards in which doctors from various fields exchange ideas about the best possible treatments. We are developing adaptive computer systems based on machine learning for certain subgroup searches of patients with particular characteristics.

Who will have access to the data?

In our consortium we have created very clear guidelines for our location: in order to guarantee data protection, the data always remain at its original location. When we receive external inquiries, we only deliver the corresponding analyses – when all participants agree to it. Nothing is automatic. Every step is manually checked.

How long will it take Freiburg to implement the plans?

I think we can achieve a lot of the action items in the next one or two years, especially the lab and medication data. We already have excellent electronic patient files. However, some of the goals won’t be reached until the end of the four-year funding period. The problem is more our lack of good medical informatics specialists in Germany. We don’t have enough required for implementing our plan at every location. Germany has missed the boat on a lot of the developments in this area, thereby shutting down a lot of Freiburg’s capacity. Medical informatics in conjunction with electronic patient files and personalized medicine is becoming more and more important.

Martin Boeker’s working group at the University of Freiburg
https://portal.uni-freiburg.de/imbi/team/persons-de/boeker-de

Medical Informatics in Research and Care in University Medicine – MIRACUM
www.miracum.de