HNU Re­search In­sti­tutes in Fo­cus - #1: Di­gi­Health In­sti­tute

Technology meets informatics meets health care: At the interfaces of this triad, which the DigiHealth team deals with, highly topical issues arise.  Today, IT systems and digital technologies have already become indispensable in the health sector, and the need for innovative solutions is constantly increasing. This involves very different dimensions of digitisation, all of which contribute to increasing the quality and efficiency of health care - classic examples are networking in the health care system (telematics infrastructure), the electronic patient file (ePA) or digital health applications (DiGA). At the DigiHealth Institute, too, the focus is on essential areas of the digitalisation of health care with mobile applications, technical assistance systems and telemedicine, but that is not all: in keeping with the interdisciplinary orientation of the Institute, research is also being conducted on the spread of diseases due to invasive insect species or the social acceptance of geothermal energy, among other things. Here, the Institute is currently planning a second doctoral position to analyse not only acceptance and feasibility but also regional needs, as Priv.-Doz. Dr Elmar Buchner reports. "Geothermal energy has enormous potential – but there is a lack of openness to technology. We want to counteract this with our research."

What factors promote or hinder the progress of digitalisation in the health sector in Germany? What are the potentials for counselling, treatment and care? How must new technologies be evaluated with regard to their effects and potentials? What is the acceptance of new technologies among employees in the health sector? These are just some of the questions that the DigiHealth Institute is currently addressing and will address in the future.
In the medium term, the focus is on securing funding through research applications, explains Prof. Dr. Walter Swoboda. Here, of course, it is also a matter of keeping the nest created for the scientific staff and – keyword temporary employment contracts – securing their jobs.
In the long term, the Institute will continue to move in the area of tension between medicine/nursing and technology: "As a comparatively small institute, we will continue to offer high-quality research, taking advantage of our high flexibility. How will AI and genetic engineering change our world? These are questions we will have to face in the future. For me personally, the ethical evaluation is particularly important."

One of the current research projects at the DigiHealth Institute is the NAP project (Neuronal Network Assisted Pathology). Here, research associate Daniel Hieber is conducting research together with DigiHealth colleagues Prof. Dr Johannes Schobel and Maximilian Karthan and the Institute of Pathology and Molecular Diagnostics at Augsburg University Hospital to improve the detection of glioblastomas – malignant brain tumours with an inauspicious, i.e. unfavourable prognosis.  The reason: the heterogeneity of glioblastomas, i.e. the question of how much the tumours differ in themselves, is, according to current knowledge, a decisive reason for their resistance to therapy: the more heterogeneous the glioblastoma, the more difficult it is to detect. 

Because the (manual) determination of this heterogeneity is a very lengthy process – a neuropathologist would be busy with such a determination for between four and eight hours – it is, however, not part of standard diagnostics.

So solutions are needed that standardise or at least simplify heterogeneity determination and thus potentially enable new therapeutic approaches – and this is where the NAP project comes in. The DigiHealth researchers are training machine learning models that automatically analyse tissue sections highlighted by haematoxylin-eosin (HE) staining. These HE slides are produced as standard in tumour diagnostics, so they do not represent an additional burden for clinical staff. With this method, which has an accuracy of 67%, heterogeneity determination in glioblastomas is possible within minutes instead of hours. 

The Institute of Neuropathology at TUM provides the necessary image data for this. The individual image scenes of the pathological sections are each around 6.5 gigabytes in size and have a resolution of 70,000 by 40,000 pixels.