Prof. Svenja Caspers (INM-1) studies the many factors that influence individual brain development. Her research focuses on how the brain ages – a process that varies enormously from person to person. “We want to understand why some people are still mentally fit at 90, while others show significant limitations at 70.”
Like her colleague Simon Eickhoff, she combines neurobiological information with other personal characteristics, drawing on vast data sets such as results from the German National Cohort (NAKO) – Germany’s largest long-term population study – and UK Biobank. At Forschungszentrum Jülich, she has also collected data as part of the 1,000 brains study (1000BRAINS), a cohort study involving around 1,300 participants, most of them aged between 65 and 85. Their brain data were supplemented with cognitive tests – for example, on memory and attention – as well as health information and lifestyle data, including place of residence, occupation, exposure to particulate matter, and alcohol and tobacco consumption. In total, the data amounted to around 90 terabytes – roughly equivalent to 90 standard computer hard drives.
“That really is big data,” says Caspers. “With a simple laptop, it would take months to process the data. That’s why we use the capacities of the Jülich Supercomputing Centre, particularly the JURECA supercomputer. Our calculations only take a few hours, a day at most.”
Here, too, machine learning methods help to identify patterns within the vast amount of data. Ultimately, the algorithm was able to identify factors that slow down or accelerate the ageing of the brain. However, when considered individually, each factor has surprisingly little influence – whether it be particulate matter, alcohol consumption, or smoking behaviour. “But we do see cumulative effects,” says Caspers.
In other words, the combination of several harmful factors causes the brain to age more rapidly, while a mix of protective influences helps to keep it young. Recommendations can be derived from this. However, these have only been of a general nature so far, explains Caspers: “We are still operating at the level of group findings. This means that the group as a whole shows a tendency, but there are individual outliers – data points that do not fit the pattern. Understanding what this means for the people behind these data points will be our next task.”
To this end, the research team will modify the methods they use. Caspers hopes that specialized deep learning algorithms could help. Another approach involves creating a digital twin of the human brain – a simplified model that depicts the network structure of the brain and can thus simulate its functional connections. “The parameters of this model could then be adjusted until they reflect the conditions of a particular person. And then it would actually be possible to make individual recommendations – why one person should exercise more, for instance, or another should avoid alcohol to help keep the brain young,” says the researcher.
And that would mark another paradigm shift in neuroscience – from the average brain to the unique, individual brain.
This text is taken from the 2/25 issue of effzett. Text: Artur Denning
