You are on page 1of 3

The Human Brain Project (HBP)

Flagship review (2016-2017)


January 2018
In the second half of 2017, the European Commission held a review of the
HBP Flagship to check on the progress of the past year. The overall
conclusion of the experts is that the project is progressing well regarding both
Science and the Research Infrastructure development. Further focus is
expected in the next phase on the user adoption of the infrastructure facilities
provided for delivering new scientific insights in neuroscience and brain
medicine.

14 high-level experts, selected for their knowledge and scientific expertise in the field, worked in
different sessions between May and October 2017 to assess the progress of the Human Brain Project
(HBP) Flagship and provide recommendations against the contract of the present project phase, covering
the period April 2016 to March 2018. It was also an opportunity for them to look at the preparations of
the Flagship's next phase that will start in April 2018.

The experts have acknowledged that the HBP teams are progressing along the planned 10 year roadmap.
In particular this advancement concerns the delivery of a research infrastructure co-designed with
scientists and industry around their different needs for neuroscience, brain medicine and future brain-
inspired computing applications. Good progress has also been made in the development of biologically-
based models which start to cover more regions of the brain thanks to the understanding gained by the
detailed mapping and mathematical modelling of new elements or circuits in the brain. This better
understanding of the functioning of the brain is validated by neuroscientists using computing
simulations, including with robots and/or neuromorphic IT implementations.

Research Infrastructure development


The experts found that the Flagship partners made good progress in further developing the six HBP
platforms1:

 The HBP Neuroinformatics platform is integrating and providing access to high resolution
atlases of rodent and human brain developed with data collected both inside and outside the HBP
Consortium. Thanks to the mining tools developed, data are now being harvested from literature

1
The HBP platforms cover neuro-informatics (including brain atlases), brain simulation (with models scaffolds),
high performance analysis and computing, medical informatics, neuromorphic computing and neurorobotics.
and new features extracted from brain images, like the 3D reconstruction of the brain structures.
Making available atlases containing functional and structural data in an integrated way is
empowering the whole neuroscience community for further analysing brain functions in both
healthy and diseased brains.
 The brain simulation platform successfully demonstrates that modelling helps understanding the
underlying functions of the brain even when not all data are available from measurements.
Greater emphasis on community adoption of such biologically-based bottom-up modelling is
now needed. In the area of high performance analytics and computing, a pilot service
architecture for providing data storage and computing has been delivered by the four 2
supercomputing (HPC) centres involved in HBP: Jülich, Zürich, Bologna, and Barcelona. The
next step is to setup a cloud-type service infrastructure3 with large data storage and high
performance computing resources dedicated to the HBP interactivity and dense memory needs
for brain simulation and data-mining.
 The HBP brain inspired Neuro-Morphic Computing (NMC) capacity is supplementing these
HPC developments. The NMC machines in Manchester (SpiNNaker-1) and Heidelberg
(BrainScaleS-1) are readily accessible and should now support collaboration with industry for
their application to neural networks, and with neuroscience to investigate the brain mechanisms
of learning (synaptic plasticity). A second generation of NMC chips is now being designed to
allow further reproduction of these dynamic processes.
 The neurorobotics platform was improved in terms of usability and features. It now enables to
test brain models both within a virtual environment (by simulating the behaviour of a robot in a
virtual environment) as well as with real robots. Focus is now needed on using this platform to
validate advanced and simplified brain models and their theoretical foundations, covering both
biological and cognitive brain functions.
 The medical informatics platform of HBP has also progressed in delivering a facility for
aggregated analysis of patients' data available in different hospitals. In the months to come,
priority will be to demonstrate the value of this platform for the discovery of biological
signatures of brain diseases using data available at hospital level. A proof of concept for
distributing data-mining across several hospitals is also foreseen. This should open new avenues
for advancing brain medicine which is essential for society.
More generally, a key success factor of HBP for the period to come is to demonstrate the value of the
Flagship's developments, by engaging further with the neuroscientific community within and beyond
HBP with specific use-cases. The intention of integrating the different HBP platforms within an 'HBP
Joint Platform' and of providing coordinated scientific and technical support to the neuroscientific users
is very promising.

Strategic Brain Data and Brain Models Scaffolding


HBP investigations of the mouse brain organisation generated foundational data that allow detailed
cellular-level brain simulations, focusing on the neocortex and using a multi-area approach for studying
the hippocampus, basal ganglia, and cerebellum. The resulting multiscale molecular, anatomical and
functional maps and models of the brain will give deeper insights for researchers and medical
practitioners necessary for better understanding neurological diseases. Next scientific challenges to be
addressed include: integrating many other dimensions of the brain like the multiple scales of the brain
(spanning from molecules to whole-brain), the long-projection neurons connecting different brain
2
The CEA computing centre is expected to join in the next phase.
3
Fenix - Federated Engine for Information eXchange.

2/3
regions, the differences between species, and the plasticity of the brain in its structures and functions.
This is important for medicine and design of new treatments (e.g. strokes, traumatic brain injuries).

HBP achieved also substantial progress on the genetic, micro and meso-architecture characterization of
the human brain and this resulted in the generation of new human brain atlases. The integration of in vivo
and in vitro experiments and computer modelling (in silico simulations) is still rare in neuroscience and
represents a unique feature of HBP. Seeking even more innovative data acquisition approaches would be
beneficial. Systems and cognitive neuroscience experimental setups, methods, data analysis tools and
simulations have resulted in new models which can contribute to better understand information
processing in the brain, which is an essential aim of HBP. Theoretical neuroscience contributed also with
other models, addressing the principles underlining the function and connection of neurons, plasticity,
and neuronal population across different scales. The model of spatial navigation should soon link with
the parallel computational simulation of circuits in the hippocampus. This brain region of paramount
importance for the memory and, in relation to the cortex, for connecting to emotions, is impaired in
several neurological diseases.

The HBP platforms are co-designed with the scientists doing brain modelling, data collections and
simulation and who need these new tools for integrating knowledge. This includes the whole mouse
brain model and the mouse brain atlas, the cortical and sub-cortical microcircuits models, the multi-level
human brain atlas and the cognitive architecture of visuomotor process of the brain. Implementation of
plasticity for learning in large-scale neuromorphic systems is starting. This will be essential for better
understanding the learning processes in the brain and also to empower future robots with brain-inspired
learning capabilities. Efforts regarding modelling for drug discovery are planned to start this year, and
hold promises for the treatment of brain injuries where synaptic plasticity is essential.

International cooperation with other brain initiatives in the world is developing notably through a
reciprocal researcher exchanges scheme as well as through regular scientific and technical dialogues in
the context of international conferences. In addition, representatives from most of the world’s major
brain research initiatives including the Brain Initiative in the USA and HBP in Europe met end 2017 and
signed a declaration to establish an International Brain Initiative to speed up progress on ‘cracking the
brain’s code' – see https://www.humanbrainproject.eu/en/follow-hbp/news/worlds-brain-initiatives-
move-forward-together/.

Finally, in the domain of Responsible Research and Innovation (RRI), the support from philosophy,
foresight analyses, citizen’s meetings and a systematic ethics management are key components well in
place for ensuring societal acceptance and ethic compliance.

About the HBP Flagship


The HBP Flagship, launched in 2013, today involves hundreds of researchers and developers in 19
countries in Europe, working to put together their scientific and medical knowledge and their
technologies and engineering capabilities. The operational phase of the created Research Infrastructure is
approaching as the Flagship is close to the mid-way point of its 10 year journey. The aim remains to
deliver knowledge to the neuroscience and other communities and facilities that are useful and
competitive at international level.

3/3

You might also like