Multiple sclerosis study at the Munich Digital Health Summit

Big Data in medicine – between progress and data protection

To bring together all medical data of a patient and to connect them with data sets of other patients, new and safe IT solutions are needed. (Image: metamorworks / istockphotos)
To bring together all medical data of a patient and to connect them with data sets of other patients, new and safe IT solutions are needed. (Image: metamorworks / istockphotos)

Research news

When pooled with other patients’ data, our personal digital medical history can help predict the course of a disease and monitor the success of its treatment more reliably. This is a great opportunity, particularly for diseases such as multiple sclerosis (MS), whose progression can vary greatly. The DIFUTURE Consortium under the direction of the Technical University of Munich (TUM) will present its work at the Munich Digital Health Summit 2018, which takes place from 29 to 30 November.

In 2015 the incidence of MS across Bavaria was 60 percent higher than just nine years before. That is the finding of a recent study conducted under the direction of Bernhard Hemmer, Professor of neurology at the TUM and member of the DIFUTURE Consortium. He and his team analyzed data from over ten million people, including nearly 30,000 MS patients in 2015.

Harnessing patient data for research

“Large medical datasets are incredibly valuable to us in the hospital. They tell us whether parallels exist in the course of the disease and whether cases share common prior conditions or clinical signs. Only with such large data pools can we draw statistically reliable conclusions that would be impossible to derive from individual patient records,” Bernhard Hemmer explains.

As part of the DIFUTURE research consortium, neurologists such as Prof. Hemmer at the TUM, Prof. Kerschensteiner at the LMU, Prof. Ulf Ziemann at the University of Tübingen and Prof. Markus Naumann and PD Dr. Antonios Bayas at Augsburg Hospital are collecting large amounts of data from multiple sclerosis patients. Together with the DIFUTURE computer scientists, they are working to unify and compile those data, which biostatisticians and bioinformaticians use to carry out analyses based on artificial intelligence (AI) and machine learning methods. Equally important is the integration of imaging data, so that neuroradiologists at the sites involved also have a key role to play.

The preliminary work, carried out with another DIFUTURE partner from the insurance sector, has already yielded initial results: MS patients are far more likely to have disorders such as anxiety, depressive episodes or nonspecific visual and emotional disorders five years before their actual diagnosis.

MS is the first disease for which the DIFUTURE scientists have developed and tested methods to harness medical data efficiently and reliably for research and clinical practice. That is a key goal of the DIFUTURE research network, which is funded by the German Ministry of Education and Research to the tune of more than 28 million euros. The new methods will also soon be applied to other disorders such as Parkinson’s disease, cancer, stroke and cardiovascular diseases.

Focus on data integration and data protection

The proper use of medical data places new demands on the collection, processing and protection of data. To be comparable and generally suitable for AI methods, data must be both retroactively standardized and, in the future, uniformly recorded. They must also be optimally protected against unauthorized access. IT and data protection experts face major technical challenges, while patients are confronted with the question of who will be permitted access to their data.

A special focus of DIFUTURE therefore lies in data protection. In distributed computing, medical data do not leave the hospital at all; they are stored solely within the hospital itself. In order to use such data along with data from other hospitals, innovative methods are at work that follow the principle of “bring the analysis to the data” (and not the data to the analysis). This is a fundamental principle behind DIFUTURE. DIFUTURE will also look at how data collected not only for research but also for healthcare can be collected both securely and without bias for research.

“More than ever before, medical science will need carefully collected and compiled data in the future. We therefore need to develop tools to ensure that as many patients as possible benefit from them. Particularly in view of the increasing complexity of networking, it must be made absolutely clear that the data belong to the individual and must be consistently protected,” says Klaus Kuhn, professor for Medical Informatics at TUM and head of the consortium.

Publications:

Data Integration for Future Medicine (DIFUTURE). Prasser F, Kohlbacher O, Mansmann U, Bauer B, Kuhn KA. Methods Inf Med. 2018;57(S 01):e57-e65. doi: 10.3414/ME17-02-0022.

A Systematic Assessment of Prevalence, Incidence and Regional Distribution of Multiple Sclerosis in Bavaria From 2006 to 2015. Daltrozzo T, Hapfelmeier A, Donnachie E, Schneider A, Hemmer B. Front Neurol. 2018;9:871. doi:10.3389/fneur.2018.00871.

Further information:

DIFUTURE is one of four consortiums in Germany set up by the Medical Informatics Initiative of the German Ministry of Education and Research. Partners of DIFUTURE are the TUM, the LMU, the University of Tübingen and their University Hospitals, The University of Saarland together with Saarland University Hospital, Regensburg University Hospital and University and Klinikum of Augsburg. The Medical Informatics Initiative aims to make optimal use of the opportunities offered by digitization in medical science for the benefit of healthcare and research.

Data from the Association of Statutory Health Insurance Physicians Bavaria (KVB) were used in the study by Prof. Hemmer.

Contact:

Prof. Dr. Klaus A. Kuhn
Technical University of Munich
TUM Chair for Medical Informatics
Institute for Medical Informatics, Statistics and Epidemiology at Klinikum rechts der Isar
Tel: +49 (0)89 4140 – 4320
klaus.kuhn@tum.de