Hospitals and researchers uncover data treasure by collaborating
Medical institutions and researchers need to collaborate to make the most of the available data. The laboratory of the University Hospital Basel is currently involved in two data network projects.
First people, then data
Big data and personalised medicine are two areas that hold great promise for health care. But progress in personalised medicine has been slow, a study of ETH Zurich published in May 2018 claims. It recognises a lack of data sharing and limited accessibility for researchers and medical institutions as the main reasons. The authors of the study add that progress will depend on collecting and sharing large amounts of molecular and clinical data as well as data on patient habits, which can then be used for various purposes and analyses.
"To achieve this aim, we first have to build functioning networks in which hospitals and researchers participate," says Adrian Egli, head of clinical microbiology at the University Hospital Basel. And that's precisely what he intends to do: setting up a network involving all university hospitals and universities. In an NRP 72 project that started in early 2018, he has brought together the labs of various hospitals and veterinary clinics to establish a central database of genetic information on pathogens. The aim is to recognise similarities between different pathogens in order to understand how they spread geographically over time. "This will enable us to track the spread of antibiotic resistance with much greater accuracy and to recognise and fight outbreaks much more rapidly," says Egli.
Only harmonised data can be compared
Adrian Egli also runs the "Personalised Swiss Sepsis Study", which is funded by the Swiss Personalized Health Network and got under way in early 2018. This study will establish infrastructures linking up the intensive care units at Swiss university hospitals and several clinical and basic research groups. The aim is to collect extensive molecular and genetic data of patients and pathogens during the entire course of sepsis. Using artificial intelligence, various research groups at ETH Zurich, led by Karsten Borgwardt, will analyse the collected data looking for patterns. They expect to identify digital and molecular biomarkers that will allow for earlier diagnosis of bacterial sepsis and more accurate predictions of the course it will take in individual patients.
Apart from their immediate scientific focus, the two projects have the same goal: they want to show that it is possible and useful to pool medical data on infectious diseases and microbiology also in Switzerland. "We are lagging far behind many other countries in this field," says Egli. One of the reasons for this is the fact that most cantonal and university hospitals have their own IT systems, that have different ways of storing and categorising data. Aitana Lebrand, who works as a clinical bioinformatician in the NRP project says: "We need to agree which data will be shared and, most importantly, define a common nomenclature that will be used to name data consistently.” This will ensure that data from different sites can be pooled and analyzed within a common bioinformatics analysis pipeline, thereby enabling comparison of data across various sites.
Informing patients, securing confidentiality
But before this work can get underway, Egli explains, they need to overcome an obstacle of a different nature: "Many institutions are reluctant to share their data. On the one hand, they worry about confidentiality, on the other hand, they are hoping to evaluate the data themselves." In terms of patient confidentiality, sharing data in large networks is reason for concern. But Egli believes that this hurdle can be overcome. Even when anonymised, these data are of great value to research. But informed consent forms and the preceding patient information would have to be reviewed before any sharing could take place.
Initially, the cooperation means additional work for institutions and researchers. Data have to be prepared according to shared guidelines and fed into a new system. But, says Egli, everyone agrees that larger datasets will allow for more representative findings. This is a major incentive. If they work together, the researchers can tackle a wide range of important questions. Even rare diseases and complications can be analysed if more data are available.
Collaborating to achieve more
"With many institutions generating more and more genomic data, a secure infrastructure with high-performance computing capacities is needed to pool and analyse these data", explains Aitana Lebrand. In the NRP 72 project, this infrastructure is provided by the Swiss Institute of Bioinformatics (SIB), where Lebrand works as a clinical bioinformatics project manager. She is in close contact with clinical partners from various institutions and the different experts at SIB who will implement a data sharing platform. Similarly, the sepsis project relies on bioinformaticians and mathematicians at ETH Zurich: the self-learning data analysis system is based on the latest software technologies.
Working together in networks does not only enlarge the evidence base, it also allows for new research approaches and methods. Every player contributing data will eventually benefit from this - in particular the individual patients, who will get a more precise diagnosis at an earlier point in time. Egli hopes that his pioneering projects will deliver in these areas: "If we are successful, we can lay the foundations of future research infrastructures which will lead to rapid progress in many areas. In this way, we can soon turn promising catchwords like big data and personalised medicine into health benefits for patients."