《歐盟（EU）：2024人類大腦計劃10年創新評估報告（英文版）（68頁）.pdf》由會員分享，可在線閱讀，更多相關《歐盟（EU）：2024人類大腦計劃10年創新評估報告（英文版）（68頁）.pdf（68頁珍藏版）》請在三個皮匠報告上搜索。

1、 Human Brain Project 10 Years Assessment EUROPEAN COMMISSION DIRECTORATE-GENERAL FOR COMMUNICATIONS NETWORKS,CONTENT AND TECHNOLOGY DIRECTORATE C-ENABLING AND EMERGING TECHNOLOGIES UNIT C.4-EMERGING AND DISRUPTIVE TECHNOLOGIES Contacts:Aymard de Touzalin,Head of Unit C4,Teresa De Martino,Senior Expe

2、rt E-mail:CNECT-C4ec.europa.eu European Commission B-1049 Brussels Directorate-General for Communications Networks,Content and Technology FP7 and H2020-Future and Emerging Technologies Human Brain Project 10 Years Assessment Members of the Evaluation Panel Gordon PIPA Michela CHIAPPALONE Hilleke HUL

3、SHOFF POL Fiona NEWELL Aureli SORIA-FRISCHch Directorate-General for Communications Networks,Content and Technology FP7 and H2020-Future and Emerging Technologies Manuscript completed in July 2024 First edition This document has been prepared for the European Commission however it reflects the views

4、 only of the authors,and the European Commission is not liable for any consequence stemming from the reuse of this publication.Luxembourg:Publications Office of the European Union,2024 European Union,2024 The reuse policy of European Commission documents is implemented by the Commission Decision 201

5、1/833/EU of 12 December 2011 on the reuse of Commission documents(OJ L 330,14.12.2011,p.39).Except otherwise noted,the reuse of this document is authorised under a Creative Commons Attribution 4.0 International(CC-BY 4.0)licence(https:/creativecommons.org/licenses/by/4.0/).This means that reuse is a

6、llowed provided appropriate credit is given and any changes are indicated.For any use or reproduction of elements that are not owned by the European Union,permission may need to be sought directly from the respective rightholders.Web ISBN 978-92-68-20722-2 doi:10.2759/6494125 KK-01-24-002-EN-N Print

7、 ISBN 978-92-68-20723-9 doi:10.2759/1520124 KK-01-24-002-EN-C Human Brain Project-10 Years Assessment Contents 1.Foreword.1 2.Executive summary.2 3.Introduction.6 3.1.Evolution of the Project in 10 years.6 4.Most Relevant Scientific Results and Impact.10 4.1.Neuroscience.12 4.2.Cognitive Science.14

8、4.3.Brain atlases.16 4.4.Neuromorphic.17 4.5.Medical Neuroscience.20 4.6.Simulation Neuroscience.22 4.7.Neuro-inspired AI.24 4.8.Robotics.25 4.9.Integration platform for digital neuroscience(EBRAINS).27 4.10.Education and Outreach.30 5.EBRAINS Research Infrastructure:present and future impact.32 6.I

9、nnovation Impact.36 6.1.Impact and Support of H2020 Goals.36 6.2.Relevance of Technology Project outputs.37 7.EU added-value.42 8.Conclusions.45 Human Brain Project-10 Years Assessment 9.Appendix.46 9.1.Evaluation methodology.46 9.2.Neurotechnology Key Innovation Areas.47 9.3.Table of Patents linked

10、 to HBP.48 9.4.Table of Companies involved in the HBP.49 9.5.Authors and Author affiliations.50 10.References.51 Human Brain Project-10 Years Assessment 1 1.Foreword The Human Brain Project flagship is one of the most groundbreaking research and innovation initiatives undertaken by the EU in the mos

11、t fascinating and complex area the human brain.Between 2013-2023,the HBP flagship brought together more than 500 scientists,engineers and clinicians from all over Europe.They shared a vision of unravelling the mysteries of the human brain through advanced computational methods and cutting-edge techn

12、ologies.This interdisciplinary collaboration resulted in a paradigm shift in brain research and in ground-breaking developments in neuroscience,computer science,robotics and data management.The HBP flagship made a remarkable progress in developing new approaches to diagnosing and treating neurologic

13、al disorders.Its strategy of using some of the worlds most powerful supercomputers for advancing neuroinformatics and data-driven research allowed its scientists to produce precise atlases of the brain,multi-scales brain models and simulations,among others.Thanks to its brain simulation and modellin

14、g algorithms,the HBP pioneered early and accurate diagnosis of brain-related conditions and personalized therapeutic strategies for neurodegenerative diseases such as epilepsy.The HBP has also demonstrated significant technological progress in brain-inspired AI and neuromorphic computing as well as

15、in cognitive robotics systems,which are of particularly high relevance for industry.Finally,the HBP succeeded in developing the unprecedented digital EBRAINS Research Infrastructure.The EBRAINS platform provides researchers with access to high-performance computing resources and advanced analytical

16、tools for sharing and analysing vast amounts of neuroscientific data.By doing so,EBRAINS has democratised brain research,enabling breakthroughs that were previously impossible.It is now used by more than 10 thousand users from more than 1500 research and medical institutes around the world.EBRAINS i

17、s now set to be the source of many more groundbreaking discoveries,in close cooperation with the planned European Brain Health Partnership,and the new Virtual Human Twin initiative.I take the opportunity to warmly thank the independent team of scientific experts,who have put together this report ass

18、essing the scientific and technological advancements and impacts of HBP.It highlights the progress achieved and looks ahead to what remains to be explored.The HBP successes of the past decade affirm that we are on the cusp of a new era in digital neuroscience,one that holds immense promise for scien

19、ce,medicine,engineering and the well-being of citizens across Europe and beyond.As we consider this potential for new discoveries,the importance of continued support and investment in digital brain research becomes ever clearer.I have no doubt that the next ten years of brain research in Europe will

20、 be just as exciting and innovative as the last decade.Roberto Viola Director General of DG Connect European Commission Human Brain Project-10 Years Assessment 2 2.Executive summary The Future and Emerging Technologies(FET)programme funded by the EU for about 30 years,supported long-term research to

21、 create new technologies,especially in multidisciplinary areas.Its mission was to turn Europes scientific strength into a competitive advantage by supporting projects that could lead to industrial leadership and solve societal challenges.Among its lines of action,the FET Flagships(or simply Flagship

22、s)were meant to support ambitious long-term,large-scale,multi-disciplinary collaborative research initiatives addressing grand scientific challenges.In October 2013 the Human Brain Project(HBP)started its adventure as one of the two first FET flagships(1)launched by the European Commission.In 2023,t

23、he HBP flagship concluded its decade-long journey,marking a successful transformation in research fields to foster a holistic understanding of the brain and the utilization of brain-inspired technologies.By 2023,the HBP flagship had engaged more than 500 international scientists in research and had

24、received EUR 607 million in EU funding.As such,this flagship project stands as one of the largest international,collaborative,and interdisciplinary research projects ever funded by the European Union.This report focuses on the HBP flagships key scientific achievements,implementation,and governance m

25、odel by conducting a comprehensive 10-year assessment of the main projects accomplishments.The specific objective of the HBP flagship consisted of developing a federated ICT infrastructure that would become a research e-infrastructure in the future,helping the neuroscience community collect,analyse,

26、share,integrate and model data about the brain with the aim of understanding better the functioning of the human brain and its diseases.With the above objective in mind,following 10 years of research,the HBP has achieved impressive results and delivered EBRAINS,an open research digital infrastructur

27、e positioned to have a transformative impact in Europe and beyond.Moreover,the EBRAINS infrastructure continues to empower new applications in brain health and brain derived technologies.The HBP flagship has established a new paradigm of digital neuroscience and a new interdisciplinary culture of co

28、llaboration.Notable achievements include leading digital brain atlases,advanced brain simulation platforms across scales,the application of cognitive modelling and personalized medicine,as well as remarkable advances in neuromorphic computing,neuro inspired robotics and AI.Considering these achievem

29、ents,the HBP has catalysed highly interdisciplinary research at scale to enhance our understanding of the brain.The legacy of the HBP flagship can be fundamental in a new phase of neuroscience,which builds upon the new research infrastructure,EBRAINS RI,in a collaborative approach involving many mor

30、e researchers across the world.Some of the identified highlights based on scientific results in the HBP are:The development of the Multilevel Human Brain Atlas,which is one of the most complete,if not the most complete,highly detailed digital 3D anatomical atlas,available on EBRAINS.(1)Flagships|Sha

31、ping Europes digital future(europa.eu)Human Brain Project-10 Years Assessment 3 The innovative Virtual Brain simulation engine,which holds great potential for modelling personalised medicine,including in an ongoing clinical trial for epilepsy treatment.Within cognitive science,diverse datasets and r

32、esearch tools have been made available on EBRAINS from a broad range of cognitive tasks for use by the wider community(e.g.Human Connectome Project Young Adult fMRI time series,structural and functional connectomes/Julich-Brain Atlas,cytoarchitectonic maps/Generative network model of visual percepti

33、on that learns invariant object representations through local minimization of prediction errors/Collaborative Brain Wave Analysis Pipeline,and more.(2)Research findings based on key cognitive functions used to constrain biologically inspired models of spatial navigation,visuo-motor interactions,and

34、dexterous manipulation were provided.Significant advancements in neuromorphic computing were achieved,potentially reshaping future computer designs for energy-efficient machines capable of brain-like learning.In particular,two hardware platforms,SpiNNaker and BrainScaleS,were transformed into open n

35、euromorphic services,thus paving the way for enhanced second-generation systems.Significant progress in AI models for applications like computer vision and robotics,offering solutions with fast reaction times and energy efficiency.Significant progress in simulation neuroscience,by integrating data a

36、cross scales,thus revolutionizing our understanding and predictive capability of brain functions.Neuro-inspired AI allowed for the exploitation of scientific results to advance machine-based technologies for AI,including interdisciplinary efforts in studying brain learning rules and to develop more

37、biologically plausible models to enhance robotics,automation,and AI applications.In robotics,advancement in embodied cognition by connecting brain models with robotic bodies was pioneered,promoting an understanding of cognition through embodied experiments.The Neuro Robotic Platform(NRP)-based innov

38、ative experiments,such as multisensory integration and precise movements,fostered research in neuro-derived AI and robotics.A versatile framework for integrating functional models was developed,enhancing research in computational neuroscience and embodied AI,and paving the way for complex cognitive

39、architectures and studies of emergent phenomena.Throughout the HBPs lifespan,the HBP platforms have evolved into the EBRAINS Research Infrastructure(RI),achieving one of the flagships primary goals.The EBRAINS RI has been developed as an integration platform for digital neuroscience and acts as an i

40、mportant research ecosystem developed by HBP.It includes over 1029 datasets,225 research software applications,160 interoperable tools and 250 models.It offers a comprehensive framework for advancing brain research,incorporating data services,model services,and a variety of digital libraries and too

41、ls,all interactively navigable through the Knowledge Graph/(in 2024:70.000 Knowledge Graph Searches and 1500-Human Brain Atlas unique users per month).(2)Find neuroscience data,models and tools on EBRAINs https:/www.ebrains.eu/data/find-data Human Brain Project-10 Years Assessment 4 At the end of th

42、e HBP,EBRAINS had approximately 2120 contributors,8475 returning users and 1491 user institutions and 11 national nodes,to make the scientific results feasible and generate this impact(based on Pioneering Digital Neuroscience,October 2023).This ecosystem enables the combination of tools for complex

43、simulation workflows,as successfully showcased in concrete use cases.EBRAINS marks an unprecedented advancement in neuroscience research,providing regulatory-compliant data access and advanced tools that enhance our understanding of the brain.(details and facts).(3&4)Overall,these achievements demon

44、strate how the scientific interest in EBRAINS has grown very considerably,especially during the last years of the flagship fostering international cooperations by partnering projects(international patterning projects were distributed across partners from the UK(20%),the US(10%),and others(with 10%to

45、gether).The inclusion of EBRAINS in the ESFRI Roadmap ensures sustainable access for researchers to state-of-the-art multi-level resources for modelling,simulation,experimentation,and data analysis.At its core,the physical hardware layer includes High Performance Computing(HPC)facilities,further dev

46、eloped within HBP and the associated ICEI project(5).This development makes the EBRAINS Research Infrastructure a one-stop-shop HPC system dedicated exclusively to advance neuroscience research and brain health applications.Furthermore,its inclusion in the Health Data Space Pilot broadens its applic

47、ation field.While the HBP flagship has prioritized infrastructure development as its main vector of innovation,it has also driven significant advances in neurotechnology.Notable progress has been made in areas such as epilepsy,prosthetics,and neuromorphic engineering.The project has produced 35 key

48、exploitable results,bridging the gap between fundamental research and innovation.Twelve of these innovations are protected by patents,with an additional 35 patents filed and awaiting decisions.These innovations are well-positioned to enter the market if effective business development support is prov

49、ided.As a direct result of HBPs efforts,the number of European neurotechnology companies has further grown,driven by the creation of 12 new start-ups.Both existing HBP private entities,whose inclusion was facilitated through the implementation of Open Calls,and new ventures are at the forefront of a

50、ddressing significant societal challenges related to demographic changes and their associated health impacts.Through the translation of innovative solutions into the brain health market,they contribute directly to European industrial leadership in the neurotechnology sector.By way of a quantitative

51、overview,the HBP flagship involved over 500 researchers from 155 institutions based across 19 countries.The HBP generated over 3000 publications,92 filed patents and 12 spin-off companies and the EBRAINS digital infrastructure for brain research.In terms of education and outreach,the HBP organized o

52、ver 500 conferences and scientific workshops and impressive number of outreach activities(i.e.58 publications for non-scientific (3)D7.9 Enabling digital neuroscience:10 years of research,co-design development,and collaboration in the Human Brain Project https:/ec.europa.eu/research/participants/doc

53、uments/downloadPublic?documentIds=080166e5056b73f3&appId=PPGMS(4)Pioneering Digital Neuroscience:how the 10-year Human Brain Project has transformed brain research hbp_pioneering_digital_neuroscience.pdf(exo.io)(5)Interactive Computing E-Infrastructure for the Human Brain Project https:/www.fz-jueli

54、ch.de/en/ias/jsc/projects/icei Human Brain Project-10 Years Assessment 5 audiences,80 exhibitions,181 videos and a large number of followers on HBP-related social media accounts including almost 40,000 followers on X and 70,000 followers on LinkedIn).It is important to note that the HBP flagship est

55、ablished 15 Key Performance Indicators(KPIs)to monitor the projects performance and EBRAINS usage.Specifically,9 out of the 15 KPIs were over-achieved(100%),particularly those related to the number of institutions,in Europe and globally,using EBRAINS and those who contributed data and models to EBRA

56、INS.In summary,the HBP significantly contributed to advance understanding of the brain and transform the related research fields to allow a more holistic approach and the use of brain-inspired technologies.Developments driven by the HBP have had a transformative impact on several brain research,and

57、technology-related fields and showed to have the potential of a transformative impact in brain health.The scientific results of the project have made an important and influential contribution towards a better understanding of the human brain.Substantial progress was made in bringing neuroscience sig

58、nificantly closer to new clinical and industrial applications.Computational neuroscience took the step towards medical applications through personalised patient brain models,which now enable the implementation of digital twin approaches.Furthermore,it served as a bridge between brain research and AI

59、,with HBP showcasing the advantages of brain-inspired algorithms,neuromorphic hardware,and neurorobotics.The basic understanding of the brain has been enriched by the multimodal and multiscale approaches facilitated by the HBP.Crucially,HBP flagship followed a rigorous programme to ensure the sustai

60、nability of its developments:it established the EBRAINS infrastructure as a lasting offering to the scientific community.Moreover,it successfully fostered a new multidisciplinary community in Europe,converging under the paradigm of“digital neuroscience”(6).(6)https:/direct.mit.edu/imag/article/doi/1

61、0.1162/imag_a_00137/120391/The-coming-decade-of-digital-brain-research-A an open community paper with more than 100 authors Human Brain Project-10 Years Assessment 6 3.Introduction In the early 2000s,a new vision shared by researchers across the world emerged that studying and understanding the huma

62、n brain,one of the most complex systems ever studied,with billions of elements across spatial and temporal scales,will require a new large-scale collaborative approach.To be successful,this approach needed to include researchers across many fields and novel technologies to model and understand the b

63、rain system.Additionally,it should involve harvesting and fusing information of breath-taking complexity.Sparked by this game-changing perspective,the European Union was the first to initiate the flagship Human Brain Project(HBP)in 2013,followed by similar initiatives from the US,Japan,China,and Sou

64、th Korea in subsequent years.Understanding the human brain implies understanding the principles underlying human cognition,intelligence,and learning.It also involves comprehending the brains capacity to self-organize,adapt,learn,and heal after injuries or trauma.Importantly,this understanding enable

65、s the treatment of brain diseases and dysfunctions,the development of new therapeutic drugs,and the construction of brain-inspired technologies to enhance artificial intelligence.This endeavour spans a wide range of disciplines,from neuroscience to medicine,computer science and AI,complex system sci

66、ence,ethics,and others.Moreover,given its relevance to individuals,society,the economy,and socioeconomics globally,a project like the HBP flagship had to maintain a broad perspective.Efforts to monitor and analyse the impacts of potential and consolidated understanding,and also to engage various sta

67、keholders,including policymakers,mass media,and the general public were central to its success.The intricate nature of these interdisciplinary fields necessitated a new level of mutual understanding,inspiration,collaboration and tool-sharing that was previously unimplemented across and within fields

68、.Today,with the conclusion of the HBP flagship and various global initiatives in their final stages,along with ongoing research efforts by the international research community,it is widely acknowledged that the human brain is still not fully understood.Understanding the human brain was,in 2013 and r

69、emains today,one of the most challenging research questions tackled by humankind.The following sections of this document provide more details following the assessment of the HBP flagship based on the achievements,implementation,and governance model,from the perspective of experts in the field who ob

70、served and evaluated the development of the HBP and its progress over the last 10 years.The assessment also focusses on the overall outcomes of the HBP,the progress made in understanding the human brain,the relevance and impact on society and the economy,and the linked innovations in the field.3.1.E

71、volution of the Project in 10 years The initial concept of HBP flagship built upon previous European funded projects on brain simulation and neuromorphic computing,such as FACETs(Fast Analog Computing with Emergent Transient States under the 6th Research Framework Program(FP)of the EU),Brain-i-Nets(

72、Novel Brain-Inspired Learning Paradigms for Large-Scale Neuronal Net-works under EU FP-7)and BrainScaleS(Brain-inspired multiscale computation in neuromorphic hybrid systems under EU FP-7)and on the Swiss Blue Brain project.The joint experience and results of these consortia led to formulating the g

73、oal of the HBP in the original grant proposal(7)with the aim“to build an integrated ICT infrastructure enabling a global collaborative effort to (7)Original Grant proposal The Human Brain Project,ICT-2013.9.9:FET Flagships.Human Brain Project-10 Years Assessment 7 address this grand challenge,and ul

74、timately to emulate the computational capabilities of the brain”.At the beginning of 2014 a debate emerged in the project and in the international neuroscience community that escalated and then culminated in July 2014 in an Open Letter to the European Commission co-signed by several hundred scientis

75、ts.The signatories requested an evaluation of HBPs governance and scientific approach and called for an independent external steering committee.In reaction to this,a mediation process was solicited by several stakeholders related to HBP flagship and formally set in motion by the HBPs Board of Direct

76、ors in September 2014.The objectives of the mediation were defined in the Terms of Reference;they included the development of a“proposal for a restructured concerted governance structure and a balanced scientific structure”.A committee of 27 members from the EU and beyond,and from inside and outside

77、 HBP,were nominated to cover a broad range of expertise in the management of scientific institutions,large-scale research projects and infrastructures and in relevant scientific disciplines.The result was a Mediation report published on March 15th,2014(8).The recommendations contained therein were r

78、elated to both the scientific profile and the overall governance of the project.The key recommendations in respect to governance were tailored to ensure openness and transparency and optimize partnership inside HBP.Key recommendations in respect to science were the following:1.define a unique set of

79、 concrete and achievable long-term objectives,which could be realized within the projected timeframe;2.focus on a smaller number of prioritized activities;3.develop a set of models complementing each other and integrating multiple scales and perspectives,together with the specification,design,implem

80、entation and testing of IT platforms enabling and exploiting these models;4.integrate cognitive and systems neurosciences into HBP;5.dedicate part of the budget to open calls to integrate new targeted research from outside HBP;6.clearly and faithfully communicate a sharpened mission statement and ob

81、jectives of HBP,which is seen as a fundamental responsibility of the consortium.The HBP consortium responded in full to these recommendations,adapted the governance structure and the lead of the project,installed new necessary bodies,and updated the scientific scope.Importantly,it included the princ

82、iple of Co-Design projects(i.e.,scientists and technologists must collaborate closely to ensure that a scientific instruments specifications are detailed enough for successful construction)to strengthen the integration of science and technology across domains,and a focus on a fewer clearly defined s

83、cientific goals.The newly implemented outreach and communication strategy effectively clarified the objectives of HBP,realigning the public perception with its original transformative goals.This shift was aimed at fostering a deeper understanding of the brain and facilitating advancements in the fie

84、ld.Despite this,the project continued to be impacted by the initial perturbations,miscommunications,or misunderstandings which occurred during the early HBP phase,and continue to cause a partial polarization in the field.(8)Mediation Report,published 15th March 2014 https:/www.fz-juelich.de/en/news/

85、archive/press-release/2015/15-10-30hbp-mediation-concluded Human Brain Project-10 Years Assessment 8 During the phases of the HBP following the mediation process,the HBP reacted continuously to new demands and improved the structure and focus of the project.It reacted with respect to the research fo

86、cus by combining a bottom-up research approach,oriented on increasing understanding from detailed nerve cells and molecular interactions,with a re-integrated top-down approach including cognitive modelling and AI.The merging of simulations with data-driven methods,and co-simulation to link scales as

87、 demonstrated in TVB-NEST(9),NRP(10),or the personalization of simulation models for clinical investigations,data-driven approaches,and linkage to high-resolution data sources like brain atlases,served to enhance predictive power.Linking the initially disparate six platforms into an integrated RI-sy

88、stem,alongside technical solutions,seamlessly merged research platforms,enabling intricate,multi-step scientific investigations.Regarding the project organization,the HBP flagship transitioned from centralized decision-making to innovative approaches that integrated diverse viewpoints,evident in bot

89、h governance and management restructuring,as well as in a number of comprehensive position papers coordinated by the Science and Infrastructure Board(SIB).A key change in the project focus was the previously mentioned re-integration of top-down research,that is research that is driven by behaviour,f

90、unctions of the brain,and conceptual models,to understand the organization of elements across scales.In particular,this topic concerned the work involved in brain simulation,where moving from many small model systems across different fields to larger data-driven models and simulations was seen as a

91、key path to better integration between the subfields of brain research.The re-integration of top-down processes accounted for the fact that scales of neuroscientific description need to be linked and that different mechanisms and principles are at work on different scales of brain organization.For t

92、his,HBP focused on enabling multiscale approaches,the need for more complex concepts,as well as organized efforts for building multidisciplinary collaborations and the appropriate technological instrumentation.During the second and the last phase of the HBP,in the years 2018-2023,the integration of

93、data-driven and simulation-based research became the key focus towards a transformative impact on the field.During the second phase,the newly introduced Co-Design projects started to produce a clear focus on science and technology and produced many highly impactful publications in the field.In the l

94、ast phase,from 2020 to 2023,the project focused on three scientific areas related to network structure and multiscale approaches that were developed.The HBP Showcases represented the final demonstrators of the joint technical developments and instrumentation,such as the atlas,simulation,and artifici

95、al neural networks,working collaboratively as a fundamental component of the primary research findings and insights generated by the project.A key element and focus of the HBP flagship was the transformation towards a new area of digital neuroscience.To this end,both the HBP consortium and the Europ

96、ean Commission enabled the research in two critical domains.Firstly,an integrated research platform(EBRAINS)(11)was developed and secondly,a computing infrastructure that provides new forms of high-performance computing resources(ICEI)(12)and new tools that enable (9)A simulation platform:https:/www

97、.humanbrainproject.eu/en/follow-hbp/news/tvb-nest-multiscale-simulation-now-available-on-human-brain-project-collab-platform/(10)Neurorobotics platform:https:/ open research infrastructure that gathers data,tools and computing facilities for brain-related research,built with interoperability at the

98、core.https:/www.ebrains.eu/(12)Interactive Computing E-Infrastructure for the Human Brain Project https:/www.fz-juelich.de/en/ias/jsc/projects/icei Human Brain Project-10 Years Assessment 9 researchers in the field to use simulation platforms,large datasets,and AI-driven analysis tools in a closed l

99、oop to match the requirements of the field.With these,the HBP has opened up new pathways between scientific fields,enabling European research to form networks and leverage new synergies between previously disparate scientific approaches.Given this success,the European Commission has launched in Janu

100、ary 2024 the EBRAINS 2.0 project granting 38 million for the further development of services within the EBRAINS research infrastructure until 2026.In the meantime,EBRAINS will continue to develop tools and services to serve the wider research communities in neurosciences,brain medicine,and brain-ins

101、pired technologies.Human Brain Project-10 Years Assessment 10 4.Most Relevant Scientific Results and Impact Given the interdisciplinary character of the HBP and its vast breadth,the assessment of the most relevant scientific results and their impact starts with the overarching Showcases,designed to

102、capitalise on the inter-disciplinary nature of the project.Following the showcases,the report is then structured into the following subtopics:Neuroscience,Cognitive Science,Neuromorphic,Medical Neuroscience,Simulation Neuroscience,Robotics,and Infrastructure developed as part of the HBP.Six Showcase

103、s,delivered in the later phase of the flagship provide examples of tangible results made by the Human Brain Project and its potential for further innovation in the future.These Showcases were further organised as follows:1 and 2 focused on the virtual ageing brain and the virtual epileptic brain usi

104、ng seizure simulations;3 and 4 focused on multiscale brain modelling of processes in different states of consciousness and multimodal perception;5 and 6 focused on modelling motor control and cognitive functions.Showcase 1,the Virtual Ageing Brain,is a mechanistic model linking changes in structural

105、 connectivity and brain function to address the inter-individual variability in decline of cognitive abilities during healthy ageing.At the core of the Virtual Ageing Brain is a dynamical brain network model informed by individual brain imaging data(structural whole-brain connectivity),and a connect

106、ivity mask selecting interhemispheric connections is used to define the age-related changes to the structure.It is hoped that this model will support future research that is needed to demonstrate how the Virtual Ageing Brain contributes to our understanding of individual ageing processes.The showcas

107、e offers future potential to integrate the theoretical and computational microcircuit,microscale network models with multiscale connectome data,which remains challenging today(Lavanga et al.,2022).Showcase 2 illustrates how improvement in epilepsy surgery is possible with the Virtual Big Brain.Indee

108、d,this showcase demonstrates the potential of digital neuroscience at the individual patient level.It uses models for realistic seizure simulations and estimation of brain parameters,showing how EBRAINS enables personalized multiscale brain simulation for improved outcomes in epilepsy surgery,valida

109、ted through the EPINOV trial(Naddaf,2023b).The results of this trial are being expected for the next months.The high-resolution and multiscale brain simulation for epilepsy and the modelling of intervention scenarios,epileptogenic zone estimation is impressive,and it could lead to improvements in cl

110、inical outcomes for patients(Jirsa et al.,2023;Wang et al.,2023).Showcase 3 was focussed on brain complexity and consciousness,and includes in its most recent setup a full brain simulation corresponding to different brain states and consciousness levels integrating meso-and micro-cortical mechanisms

111、.This showcase was based on different actions of anaesthetics and compared to the awake condition.The level of synchrony between three species,mouse,monkey and human was compared,and differences identified due to axonal signal propagation between regions associated with brain size differences betwee

112、n the species(Sacha et al.,2024).The implementation was done in The Virtual Brain to yield models where the action on synaptic receptors can be evaluated at the large-scale.Human Brain Project-10 Years Assessment 11 Showcase 4 consisted of simulations of object and scene recognition across scales,in

113、cluding a detailed cortical column network,introducing synaptic plasticity and rhythmic oscillations,as well as showing predictive coding at the cellular level based on a cortical column model,and is based on the WhiskEye robot.The models are based in EBRAINS and present a plausible,cognitive archit

114、ecture that can be learned(Pearson et al.,2021).Showcase 5 has as its primary focus the simulation of in-hand object manipulation performed by an anthropomorphic model,the Shadow Dexterous Hand.This training was carried out in silico.It realised a hierarchical and modular active inference model of t

115、he hippocampal prefrontal circuit that addresses memory guided spatial alternation tasks,by learning and then combining cognitive maps of both physical space and task space(Van de Maele et al.,2023).Showcase 6 demonstrates a closed-loop simulation of advanced cognitive and sensorimotor functions.It

116、focuses on motor control encompassing cerebellar and spinal networks,with parallel fibre plasticity in the cerebellar module and reflex circuitry in the spinal cord.Various levels of model implementation are described from biologically inspired hardware(Bruel et al.,2024).These showcases are a resul

117、t of the agile management approach that was pioneered by HBP and can be a blueprint for future large-scale joint research in fast innovating fields.The agile management approach,meticulously fine-tuned by HBP,has provided a much-needed framework for effectively managing and executing complex researc

118、h endeavors.By embracing agility,HBP has successfully navigated the dynamic landscape of fast-paced innovation and seamlessly integrated diverse research efforts.This groundbreaking strategy not only enables the assimilation of ideas and findings from different disciplines but also fosters a collabo

119、rative environment that accelerates progress towards coveted breakthroughs.Indeed,HBP has pioneered the Virtual Brain multiscale implementations,linking biological realism to stimulated brain activity.The showcases developing biologically informed computer models for motor and cognitive functioning

120、have revealed the complexity of such tasks(Sacha et al.,2024):bridging levels from the microscale in the brain up to behaviour is still in its early phase although significant advances were made.Importantly,information is available to construct large-scale models although simulating such models at t

121、he cellular scale represents a huge investment of computational resources and is out of reach for most researchers with no access to such resources.An alternative approach is to simulate brain activity using population models,which are much less demanding on computational resources.However,such mode

122、ls need to contain enough biological realism to include the relevant biophysical mechanisms necessary to generate brain states,such as membrane conductance and synaptic receptor types(Sacha et al.,2024).The future will learn how these brain models can inform brain changes up to the cognitive and beh

123、avioural levels in the individual.To date,however,the HBP has shown the potential for bridging the brains scales using digital models that are biologically informed.Human Brain Project-10 Years Assessment 12 4.1.Neuroscience Neuroscience has played a central role within the Human Brain Project,focus

124、ing on the study of the nervous system and brain.The project has made significant contributions to both the understanding of brain structures and their functions,as well as their potential integration.In terms of structure,the project has developed a highly detailed digital 3D anatomical atlas,known

125、 as the Multilevel Human Brain Atlas.This atlas is made accessible on the EBRAINS platform,enabling users to navigate through it,utilize it for modelling purposes,and connect it with other resources and atlases.In terms of brain function,the project has made strides in the development of digital tec

126、hnologies that can simulate human brain functions using virtual models.One key application is the Virtual Brain simulation engine for epilepsy,known as the Virtual Epileptic Patient.This cutting-edge technology is currently undergoing clinical trials,making it the first clinical neuroscience applica

127、tion of its kind.A successful validation of the Virtual Epileptic Patient holds immense potential in facilitating the development of personalized medicine approaches and improving clinical outcomes.Moreover,HBP has advanced and pioneered on many levels the integration of anatomical and functional da

128、ta through multiscale modelling at the cellular level.This approach shows convincing and converging benefits that warrant further development.By establishing effective links between anatomical and functional information,the project aims to deepen our understanding of the brain and enhance our abilit

129、y to comprehend its complexities.The Multilevel Human Brain Atlas is a highly detailed,digital 3D anatomical atlas that serves as a fundamental resource of the Human Brain Project.The atlas is built on the principles of human cytoarchitecture,a microstructural parcellation of the brain.The Multileve

130、l Human Brain Atlas consists of maps of post-mortem brain sections of 20 micrometre thickness stained for cell bodies and digitally imaged.Digital imaging of the stained brain sections combined created a three-dimensional atlas containing cytoarchitectonic maps of cortical areas and subcortical nucl

131、ei.Multimodal data on fibre and receptor architecture as well as function is connected to these maps.The cytoarchitectural basis of the Multilevel Human Brain Atlas is also called the Julich Brain Atlas(Amunts et al.,2020).The atlas is probabilistic,based on the brains of multiple individuals,which

132、enables it to account for variations between individual brains.Currently,the Multilevel Human Brain Atlas contains over 200 probabilistic maps.Moreover,it contains the so-called BigBrain(Amunts et al.,2013),an ultrahigh-resolution reconstruction of an individual brain and microscopical reference spa

133、ce at nearly cellular level that informs,e.g.,Showcase 2.The spatial resolution of the BigBrain space is unique,and allows to integrated data from micro to macro,thereby addressing the need to approach the human brain as a multi-scale system.Recently,the atlas was enhanced by the integration of fibr

134、e architecture within the human hippocampus,using three-dimensional polarized light imaging(3D-PLI)in post-mortem tissue(Axer&Amunts,2022).This integration allowed for detailed connectivity Highlights,Outcomes and Impact Excellent progress in understanding of brain structures and their functions Dev

135、elopment of the Multilevel Human Brain Atlas,a highly detailed digital 3D anatomical atlas,that is accessible on the EBRAINS platform A ground-breaking technology,the Virtual Brain simulation engine,which holds great potential for personalized medicine improving clinical outcomes Human Brain Project

136、-10 Years Assessment 13 information to be combined with cytoarchitecture information.Additionally,brain connectivity data from in-vivo diffusion magnetic resonance imaging(dMRI)of 78 healthy individuals was incorporated,providing comprehensive fibre connectivity coverage across the entire brain.The

137、integration of dMRI data with the Multilevel Human Brain Atlas was optimized based on data from post-mortem hippocampus tissue(Axer&Amunts,2022),which paves the way for potential integration with other in-vivo brain imaging data in the future.Recent analysis using the Multilevel Human Brain Atlas,in

138、 combination with studies on neurotransmitter receptor genes and receptor densities in auditory,somatosensory,visual,and motor systems,has revealed covariation and specific gene expression patterns in functional systems(Zachlod et al.,2022).This combined analysis demonstrates the potential for the a

139、tlas to provide new insights into human brain structure and potential brain functioning.Virtual Brain models have been developed for several applications,including simulations to gain better insight into individual differences in brain ageing,learning,Parkinsons disease,Amyotrophic Lateral Sclerosis

140、,Alzheimers Disease,depression,and epilepsy.Models are being informed by data from the Multilevel Human Brain Atlas.These Virtual Brain models have demonstrated their potential in early work,subject to continuous improvement with the potential to fully demonstrate their future impact.Standing out is

141、 the work on epilepsy,which provides an important showcase for bridging the gap between fundamental research and innovation up to the level of potential novel interventions of the Human Brain Project.Here,virtual brain models have been developed of epilepsy patients who do not respond to medication.

142、The virtual brain model is based on a patient-specific whole brain network model that is constructed from anatomical T1 and diffusion-weighted magnetic resonance imaging(Wang et al.,2023).Each network node is equipped with a mathematical dynamical model to simulate seizure activity.Bayesian inferenc

143、e methods sample and optimize key parameters of the personalized model using functional stereo-electroencephalography recordings of patients seizures.These key parameters and their personalized model determine the epileptogenic zone networks of the patient.The virtual brain models help to identify t

144、he brain areas where seizures emerge and can be considered as a digital twin.In health research,digital twins can be defined as virtual representations of patients that are generated from multimodal patient data,population data,and real-time updates on patient and environmental variables(Venkatesh e

145、t al.,2022).Importantly,the Multiscale Human Brain Atlas holds promise for further integration with human in-vivo magnetic resonance imaging endeavours developing at even higher field strength,with the breath of existing data available worldwide and developed in parallel to HBP,such as in ABCD(Karch

146、er&Barch,2021),ENIGMA(Thompson et al.,2020),HCP(Van Essen et al.,2013),and UKBiobank(Sudlow et al.,2015)among others,and with clinical data.Also,the Multiscale Human Brain Atlas can provide the basis on which important plastic changes taking place in the brain can be investigated using longitudinal

147、data to obtain models for development and ageing.Furthermore,the theoretical and computational microcircuit,microscale network models have potential to integrate with multiscale connectome data,which is challenging but an exciting opportunity.The Virtual Brain models successfully predict the evoluti

148、on of brain activity within the anatomical constraints and can be personalized to describe changes in a participants brain activity across brain states and potentially across a range of diseases.The models currently lack the capacity to represent functionally meaningful processes such as cognition,a

149、nd while the prototypes draw inspiration from visual and auditory discrimination tasks,so far,they offer the potential for experimental validation and deeper insights into underlying brain mechanisms.Taken together,neuroscience in the Human Brain Project has made significant,relevant,and impactful p

150、rogress towards a common structural brain research reference framework in the EBRAINS infrastructure,based on a highly detailed anatomical core.It has an inbuild potential Human Brain Project-10 Years Assessment 14 to improve even further as more data become available.Moreover,the HBP has successful

151、ly demonstrated that integrating multiple scales and virtual brain simulations are feasible,add value,and require building new bridges.The virtual brain simulations hold promise for clinical applications.Indeed,these developments have significantly increased the capacity of the neuroscience communit

152、y to model multiscale neural structure and functions of human brain networks,sustainable for the future in the EBRAINS infrastructure.4.2.Cognitive Science One of the main achievements of the HBP was to attempt to fill an important gap in neuroscience:bridging scales to allow understanding of the hu

153、man brain at all levels,from cellular,microcircuits and system-based mechanisms and architectures,leading to a better understanding of human behaviour,thought and cognition.In turn,this knowledge would drive the potential/industry impact of neuro-inspired event-based AI and robotics.The concept behi

154、nd this approach was sound,and overarching aim and related objectives were clear.However,the ultimate goal of elucidating human behaviour and cognition through neuroscience was not sufficiently supported within the project particularly in the early stages,although the influence of this work grew tow

155、ards the latter stages of the HBP.Throughout,striking the right balance between scientific research of human brain and behaviour,and implementing this knowledge in the development of neuro-inspired models and building of necessary ICT infrastructure,was always challenging.At the early stages of the

156、project,activities focused on cognition were not well integrated into the HBP despite their emphasis within the project(i.e.the overall project was structured around 12 sub-projects,4 of which were mainly based on scientific outputs).Furthermore,the abrupt change in the research groups focusing on c

157、ognitive processes during the project(2016,following mediation as discussed above)had an impact on the cohesiveness and integration of research on human cognition into the overall project.An early decision was made to restructure the work subsumed under Cognitive Architectures meaning it was subsequ

158、ently scaled back and the relative allocation of resources(budget/personnel)was reduced.Although to some extent this compounded the issue at the time,ultimately however,it provided the opportunity to focus on fewer core cognitive functions in more depth.An open call was initiated directly after this

159、 restructuring resulting in submissions that were of very high quality and the project was eventually steered around four main topics that captured a broad range of essential cognitive abilities.Thereafter,the HBP consortium demonstrated that they could successfully bridge the gap between neuroscien

160、ce,cognition and the development of neuro-inspired ITC technology.At the initial stages of the project,the emphasis was on a selection of key cognitive processes.These included:Perception and Action;Motivation,decision and reward;learning and Highlights,Outcomes and Impact Significant advances in ou

161、r understanding of human cognition and its brain bases,realized in a large number of publications in leading international journals.Availability of large,diverse datasets and research tools made available on EBRAINS from a broad range of cognitive tasks for use by wider community.Provision of resear

162、ch findings based on key cognitive functions used to constrain biologically inspired models of spatial navigation,visuo-motor interactions,and dexterous manipulation.Human Brain Project-10 Years Assessment 15 memory;space,time and numbers;and multimodal processing.In addition,efforts to characterize

163、 cognitive functions unique to the human brain were also conducted using comparative studies.Whilst each of these projects resulted in significant contributions to knowledge in their own right e.g.leading to a Special Issue of Neuron(Dehaene et al.,2015),and these projects leveraged some important c

164、ollaborations between researchers working on similar issues,efforts at integration into the wider overall goals of the HBP during this early phase of the project were at best,only partial:the projects were not(yet)sufficiently linked with the pipeline of work from brain models(including building of

165、the brain maps and Theoretical Neuroscience)to ICT such as Neurorobotics Platform.Moreover,the scientific investigations relating to cognition and the brain proceeded mainly independently,often without capitalizing on significant advances on connectomics and neurodevelopmental research findings from

166、 outside of the HBP.Nevertheless,these activities contributed towards an important legacy of the HBP project,namely the provision of data sets and experimental protocols from behavioural and related neuroimaging(fMRI and EEG)studies of the human brain as well as a considerable contribution to scient

167、ific knowledge in terms of journal article publications(i.e.almost 150 under the theme of on Understanding Cognition alone,as sourced on the HBP website).Whilst data sets were likely useful to the wider community the subsequent usage of these contributions by other sections within the HBP consortium

168、 was unclear.At the early stages of the project in particular,although the activities conformed to proposed work outlined in the grant,the findings relating to human cognitive functions and underlying neural correlates rarely went beyond the state-of-the-art in that there were no major breakthroughs

169、,mainly as a consequence of failing to capitalise on the rich infrastructure and broad resources available in the project.The latter phases of the project attempted to rectify this lack of integration by focusing on fewer but core cognitive functions that had a more obvious trajectory within the goa

170、ls of the overall project,particularly through Co-Design Projects(a good example is CDP4 which focused on visuomotor integration).This focus helped demonstrate a more direct link between the outputs of basic science to the development of technology and as an overall strategy it proved to be much mor

171、e successful than previous efforts(Bjerke et al.,2018).These functions ranged from states of conscious awareness(Salomon et al.,2016;Storm et al.,2017;Aubinet et al.,2018;Pennartz,2018;Annen et al.,2019;Martens et al.,2019;Suzuki&Larkum,2020;Pennartz,2022),sleep(Capone et al.,2019;Pastorelli et al.,

172、2019)perceptual processes(Gerard-Mercier et al.,2016;Edwards et al.,2017;Spoerer et al.,2017),multisensory processing(Meijer et al.,2017;Meijer et al.,2018;Oude Lohuis et al.,2022),spatial(Chen et al.,2018;Chen et al.,2019),episodic(Prescott et al.,2019)and procedural(Hesseg et al.,2016)memory.Furth

173、ermore,the consortium adopted realistic but ambitious proposals to build models that were biologically and behaviourally inspired.Importantly,during the interim and latter stages of the project a remarkable amount of new and important data,methods,tools,maps and atlases was generated(although the tr

174、ansition of scientific data to EBRAINS was still ongoing by the end of the project).This led to new scientific insights into inter-subject(Zhou et al.,2016)and cross species variability of the brain(from tadpoles,e.g.(Terni et al.,2017),to mice e.g.(Meijer et al.,2017),ferret e.g.(Klaver et al.,2023

175、),non-human primates,e.g.(Yao&Vanduffel,2022)as well as variability in humans(Amunts et al.,2020;Lavanga et al.,2022).Moreover,there was a demonstrable increase in the influence of these scientific efforts on other developments across the HBP,including multiscale models(i.e.multisensory deep predict

176、ive coding model that is neurobiologically inspired),simulations and information for robotic implementations of learning,memory,and multisensory integration in the Neurorobotic platform(Knoll et al.,2016).Beyond these achievements within the HBP the impact of this work was also realised in many scie

177、ntific publications,including in leading international journals.Examples of successful integration of Human Brain Project-10 Years Assessment 16 cognitive models into the project include the applications of predictive coding and analysis of the role of context in processing low-contrast visual infor

178、mation to understand object perception and memory(Showcase 4),and motor-coordination of hand movements for implementation in a robot model(Showcase 5).Other examples,such as visual prosthesis for sight restoration(Chen et al.,2020),have the potential to capitalise on EBRAINS infrastructure in the fu

179、ture to develop technology for future human-specific requirements.However,either due to missed opportunities available within the overall project or due to limited resources,there still remained some(interesting)scientific investigations that appeared to lie outside the immediate goals of the HBP(or

180、 their implementation in modelling or simulations was unclear),even at the late stages of the project.Thus,it was often unclear how the HBP benefitted directly from these stand-alone or fragmented scientific investigations,irrespective of their quality.Some examples include the role of feedback sign

181、als on sensory and decision-making processes in the brain;olfactory pathways;personality traits;cognitive decline in age-related disorders including kidney disease;environmental stressors and human brain function;and numerosity and face perception.These research threads were,nevertheless,timely and

182、of interest to the community ensuring their impact on fundamental knowledge through scientific publications(Lee et al.,2023;Martial et al.,2023;Petro et al.,2023)Furthermore,although data curation is ongoing,the future availability of these(and other)respective datasets on EBRAINS will help ensure l

183、ong-term impact.4.3.Brain atlases The human brain is organized across many spatial scales.Brain organisation is being approached by a multitude of studies carried out by different specialized scientific disciplines.However,results coming from all these studies in the end reflect the interaction of o

184、ne organ,the brain.To address this challenge and to advance our understanding thus means to better integrate our insights into the coherent framework of a multilevel atlas.The HBP has developed a new kind of atlas,a kind of google maps.But for the brain,that goes beyond the interest of a single subc

185、ommunity,and addresses the multi-level brain organization in a comprehensive way.The rich toolbox and new strategies developed for this multi-scale era of brain research are available on EBRAINS in the form of high-resolution multi-scale atlases of different species the human brain,the rat brain and

186、,in the last phase of the HBP,also the monkey brain.This is Highlights,Outcomes and Impact Taking the lead in developing and providing a new kind of multilevel atlas including brain maps for different species within the same logic and framework,that allow to seamlessly connect the data from the atla

187、s to modelling and simulation,bigdata analytics and other applications Developing and realizing a comprehensive tool suite of software including siibra,QUINT and other tools for handling heterogeneous,complex and big data Significant progress in mapping the brain at all levels of organisation,with m

188、icrostructure as a key reference,to enable data integration at both higher scales(e.g.,neuroimaging),and lower scales(e.g.,high-resolution optical methods,which make the atlas one of the most popular tools of EBRAINS Human Brain Project-10 Years Assessment 17 linked to computational models,reproduci

189、ble workflows and advanced digital tools that can be applied across multiple scales.The human brain atlas(Amunts et al.,2020)provides different spatial templates,to address to the need of requirements coming from different research questions.While the BigBrain model provides a nearly cellular resolu

190、tion,two MNI-reference spaces respond to the need of the neuroimaging community,and provide comprehensive maps of fiber tracts(Guevara et al.,2022)and functional parcellations(Thirion et al.,2024).Importantly,all references are interoperable.Moreover,the concept foresees intersubject variability as

191、key feature of brain organization,and provides probabilistic,quantitative maps on brain architecture and function.The project has delivered the first extensive open access 3D rat brain atlas,the Waxholm Space atlas of the rat brain(Kleven et al.,2023).This atlas has become a fundamental resource in

192、the neuroscience community,facilitating data integration when used in combination with EBRAINS tools for registering 2D and 3D data to atlases.It provides standardization of spatial data from the rat brain and supports a wide range of research,as well as educational and technological applications.Co

193、mbined with analytical workflows such as QUINT,it offers a robust framework for integrating and analyzing diverse neuroanatomical data(Leergaard&Bjaalie,2022).4.4.Neuromorphic Achieving simulations of large-scale brain networks up to the cellular resolution,both in space and time,demands the use of

194、supercomputers.However,this requirement comes with a significant need for both time and energy consumption.Todays supercomputers typically require several minutes to simulate just one second of biological time and consume substantial amounts of kilo(KW)or even megawatt(MW)of power.Consequently,resea

195、rch involving processes such as plasticity,learning,and development,which unfold over hours and days of biological time,remain beyond current ICT capabilities.The HBP has very successfully addressed the efficiency concern,potentially changing the future perception and design of computers.The identif

196、ied solution relies on Neuromorphic Computing,an interdisciplinary field which involves the development of hardware chips designed to function similarly to biological neural networks.Thanks to this peculiarity,neuromorphic-based systems present key features,typical of the living brains,such as paral

197、lelism,low power consumption,adaptability and fault tolerance,making them the perfect candidates to overcome conventional von Neumann computer architectures(Hoefflinger,2011).Indeed,hardware implementations of neuronal networks offer the potential to conduct simulations within a timeframe comparable

198、 to or shorter than biological time,all while consuming low levels of power.Highlights,Outcomes and Impact Pioneering advancement in neuromorphic computing,potentially reshaping future computer designs for energy-efficient machines capable of brain-like learning Two hardware platforms,SpiNNaker and

199、BrainScaleS,transformed into open neuromorphic services,paving the way for enhanced second-generation systems Significant progress in AI models for applications like computer vision and robotics,offering solutions with fast reaction times and energy efficiency Human Brain Project-10 Years Assessment

200、 18 In the original proposal,the HBP aimed to exploit neuromorphic principles to construct more efficient machines able to evolve and learn like the brain,thus overcoming one of the major barriers in the advancement of ICT systems.Indeed,the main focus was on building systems for brain sciences,espe

201、cially to model brain subsystems and functions.During these 10 years,there was a significant surge in industrial AI,which occurred parallel to the development of the HBP,thus the interest in neuromorphic-based applications shifted more towards AI.Most of the AI work initially started to be carried o

202、ut by means of GPUs,as at the beginning of the HBP it was not so clear that neuromorphic could support low energy/large scale mainstream AI.By the end of the HBP,a lot more interest in the potential of neuromorphic systems to address some of the issues in mainstream AI was developed.Major identified

203、 applications included technical assistance to humans,real-time diagnostics of complex machinery,autonomous navigation,self-repair,and health monitoring.To achieve its challenging objective,the HBP could rely on some of the major European key players in Neuromorphic Computing,previously involved in

204、a series of projects,such as FACETs,BrainScaleS,Brain-i-nets,ECHORD,SpiNNaker.Moreover,the project benefitted from the expertise of a member of the Science and Infrastructure Advisory Board(SIAB)of the HBP,namely Prof.Giacomo Indiveri from the University of Zurich,who is one of the worlds leading sc

205、ientists in the field.The HBPs work in Neuromorphic Computing specifically capitalized on two of the above cited projects:SpiNNAker(13)(Painkras et al.,2013)and BrainScaleS(14).They are quite different but complement each other well in how they work.SpiNNaker was born as a UK funded research project

206、 whose goal was to build neuromorphic computing systems based on many core chips with efficient bi-directional links for asynchronous spike-based communication(15).The SpiNNaker platform is a digital system featuring a multiprocessor architecture composed of ARM processors.SpiNNaker achieves real-ti

207、me performance with an integration time step of 1ms,which suits very well applications in robotics and artificial neural networks.BrainScaleS(16)started as an EU-funded research project integrating in vivo experiments with computational analysis to investigate how the brain processes information on

208、multiple spatial and temporal scales,to implement these capabilities in neuromorphic technology.The BrainScaleS system comprises a series of modules,each consisting of a wafer that integrates 448 analog neuromorphic chips alongside a routing system.Within each module,the emulation of 512 neurons and

209、 115,000 synapses takes place,achieving computational speeds 104 times faster than biological time.Even if both Spinnaker and BrainScaleS existed before the HBP,having undergone their original design and development in other projects,the HBP played a crucial role in supporting the significant softwa

210、re activities needed to transform these pre-existing hardware platforms into open neuromorphic services.This involved the construction of software stacks for both machines and the development of the PyNN language(Davison et al.,2009),which was utilized for both Spinnaker and BrainScaleS.Then,the pri

211、mary achievement of the HBP in neuromorphic computing lies in the conversion of these systems into services.Additionally,the HBP supported the development of second-generation for both platforms,resulting in the emergence of Spinnaker2 and BrainScaleS2,which was also integrated into EBRAINS(Billaude

212、lle et al.,2020),towards the end of the project.While these second-generation systems did not feature prominently in the immediate results of HBP,they hold significant promise for future developments,thanks to their unique combination of versatility and speed in emulating spiking networks.Notably,Sp

213、iNNaker2 is (13)https:/www.humanbrainproject.eu/en/collaborate-hbp/innovation-industry/technology-catalogue/spinnaker/(14)https:/www.humanbrainproject.eu/en/science-development/focus-areas/neuromorphic-computing/(15)https:/apt.cs.manchester.ac.uk/projects/SpiNNaker/(16)https:/brainscales.kip.uni-hei

214、delberg.de/index.html Human Brain Project-10 Years Assessment 19 being commercialized through the Technical University of Dresden spin-off company SpiNNcloud Systems GmbH(17)thanks to national grants and industry support from prominent companies like BMW and Infineon,with whom it collaborates on pil

215、ot projects to investigate the exploitation of neuromorphic computing in their respective business sectors.BrainScaleS2,on the other hand,contributed to advancements in learning operations and expanded the systems functionality,including dendritic branch modelling.These enhancements would not have b

216、een feasible in the initial release of the systems.Along with Intels Loihi technology(18),these are the only large-scale neuromorphic systems in operation today anywhere in the world,but Spinnaker and BrainScaleS remained the only publicly available,openly accessible neuromorphic systems worldwide,r

217、endering them highly unique.Among them,Spinnaker stood out as the largest neuromorphic computing platform in terms of its capacity to handle biological-scale computations(at least until April 2024).Notwithstanding,it is important to underline that Intel has not focused on brain modelling,which was,i

218、nstead,the main objective of the HBP in neuromorphic development.Within the HBP,there existed a synergistic cooperation among various groups to attempt the implementation of cortical microcircuits,originally developed by Julich.Different groups were pursuing various types of implementations:some on

219、supercomputers,some on neuromorphic platforms,and others on GPUs.Their aim was to achieve biological real-time processing.Spinnaker was the first to achieve important results(Rhodes et al.,2020),albeit based on older technology.Subsequently,all the other groups achieved similar results but using muc

220、h more recent technology.Both Spinnaker and BrainScaleS possessed the capability to seamlessly integrate software to support new learning rules(Bellec et al.,2020;Cramer et al.,2020;Stckl&Maass,2021).A notable example is the eProp algorithm(Rostami et al.,2022).eProp is an algorithm that effectively

221、 implements graded descent learning,achieving a biologically realistic management of time.In terms of applications,thanks to HBPs outcomes in neuromorphic computing,advancements have been made in impactful AI models for computer vision(Gltz et al.,2021;Zenke et al.,2021),quantum tomography(Czischek

222、et al.,2022;Klassert et al.,2022),robotics,industry,and autonomous systems,where fast reaction times,low latency and energy efficiency are essential(Yik et al.,2023).The HBP strongly encouraged and supported the collaboration between theoreticians,modelers,hardware,and software developers to signifi

223、cantly improve both Spinnaker and BrainScaleS.Being part of the HBP significantly facilitated the process of finding individuals with similar interests and ideas.In this regard,it presented a remarkable opportunity for all the teams involved in advancing neuromorphic technology.(17)https:/ Human Bra

224、in Project-10 Years Assessment 20 4.5.Medical Neuroscience Today,almost one in three people globally will develop a neurological disorder at some point in their lifetime(Feigin et al.,2020).Moreover,the Global Burden of Diseases,Injuries,and Risk Factors Study(GBD)2019 shows that mental disorders re

225、main among the top ten leading causes of burden worldwide,with no evidence of global reduction in the burden since 1990.The total European(in 2010)costs of brain disorders was 798 billion Euros(Olesen et al.,2012).Indeed,it was recently stated in a WHO Position paper(2022)that optimizing brain healt

226、h improves mental and physical health and creates positive social and economic impacts,all of which contribute to greater well-being and help advance society.The Human Brain Project approached this challenge of understanding the functioning of the human brain and its diseases by studying the brain a

227、t multiple levels(in parallel)while trying to bind these levels together because,in the words of director of the HBP,Katrin Amunts,“Research on the brain requires an understanding of the multilevel and multiscale of the brain”(Naddaf,2023a).In medical neuroscience,the focus of the Human Brain Projec

228、t has been on personalized health and diseases of the human brain.In the first years of the Human Brain Project,efforts went into building an international platform for sharing clinical data,the Medical Informatic Platform(MIP).The Medical Informatics Platform(MIP)was developed to provide a cloud ba

229、sed,federated,open-source service for advanced analytics for diagnosis and research in clinical neuroscience.The early development and wide aim of the MIP exposed multiple challenges that needed to be overcome,including ensuring secure data-sharing of multicentre clinical data internationally while

230、taking GDPR regulations into account.The federated aspect of the MIP is laudable since it potentially allows data to stay locally and moves around software for analysing the data across nodes.However,the use of the MIP has remained limited.More recent efforts for data sharing,such as the Human Intra

231、cerebral EEG Platform(HIP)and EBRAINS Health Data Cloud started at a smaller scale and are more targeted to specific projects,which seems to work better.Importantly,in recent years,the focus was more on developing digital brain models to improving brain health,such as in epilepsy and major depressiv

232、e disorder,and for a good reason.The aim of better understanding the functioning of the human brain and its diseases remains as relevant today as it was at the start of the Human Brain Project,despite the significant contributions made toward this goal.Indeed,substantial progress has been made in th

233、e Human Brain Project in developing digital models to better understand the functioning of the human brain and its diseases;the future will learn how it will lead to improvement for clinical medicine to optimise brain health.A particular highlight was the work on epilepsy,which provides an important

234、 showcase for bridging the gap between fundamental research and innovation up to the level of potential novel interventions of the Human Brain Project.Here,virtual brain models have been developed of epilepsy patients who do not respond to medication.Currently,a clinical trial using the virtual High

235、lights,Outcomes and Impact The Virtual Brain models showcase bridging the gap between fundamental digital brain research and innovation up to the level of a clinical trial in epilepsy patients The potential for detailed brain maps to aid personal treatment of brain diseases The importance of digital

236、 platforms for sharing data across borders to advance science for brain health Human Brain Project-10 Years Assessment 21 brain models is ongoing in an expected 356-patients with the aim to provide surgeons with a precise tool to help individual surgery decisions and improve outcomes(Human Brain Pro

237、ject Task Force for Science Communication,2023;Wang et al.,2023).Indeed,with improvement of predictive power of personalized virtual brains models,and with further testing in clinical trials,virtual brains may inform clinical practice in the near future(Jirsa et al.,2023).This work is impressive and

238、 can be considered as an example of the potential of digital twins to medical neuroscience and personalised medicine.The potential of including novel microstructural maps of the human brain in future research in the field of psychiatry was shown in a study on major depressive disorders.Subregions of

239、 the frontal poles of 73 depressed patients and 73 healthy individuals were compared using structural magnetic resonance imaging(Bludau et al,2017).This was combined with cytoarchitectonic maps of the human frontal pole showing two distinct areas within the frontal pole,and by statistical learning a

240、lgorithms it was found that the left medial frontal pole has the most discriminative morphological and genetic features of the frontal pole subregions.Including cytoarchitectonically specific maps(Julich-Brain Atlas)in the current analytic framework allowed for assessing the spatial arrangement and

241、the degree of volume loss in the frontal pole in depressive patients.Indeed,these microstructural maps and the JuGex tools on EBRAINS allow for a more specific assessment of regional atrophy than is possible by macro-anatomical definitions(Bludau et al.,2016)and to identify differential functional i

242、nvolvements and genetic profiles of brain areas(Bludau et al.,2018).Recently,unrelated to the Human Brain Project,the utility was demonstrated of the first objective brain-based biomarkers in the management of personalized deep-brain stimulation of the subcallosal cingulate for symptom relief for tr

243、eatment-resistant depression(Alagapan et al.,2023).Detailed brain maps such as developed in the Human Brain Project may hold the potential to aid such personalized treatment of severe psychiatric disorders in the future.Research and clinical advancements in personalized medicine require large-scale

244、data sharing to ensure progress.The Human Brain Project aimed to provide services for advanced analytics for diagnosis and research in clinical neuroscience to ensure that clinical data could be shared in a GDPR compliant manner.This has led to the development of three platforms offering services on

245、 an international scale within a GDPR-compliant,cloud-based,trusted research environment(TRE):The Medical Informatics Platform(MIP),the Human Intracerebral EEG Platform(HIP),and more recently the EBRAINS HealthDataCloud(EHDC).Services for advanced analytics for diagnosis and research in clinical neu

246、roscience are highly needed by science and medicine.However,building such services has not been an easy task.With a personalized clinical medicine scope,focusing on the individual,such platforms are rather complex to implement,and overall progress has been slow.Currently,the MIP includes a federated

247、 network of multiple distinct European centres and hosts several disease-related models in the fields of epilepsy,stroke,dementia,traumatic brain injury and mental health(19).The HIP is smaller,focused on a specific kind of data and clinical trial,and more centralized,making it easier to operate aft

248、er compliance for data sharing is provided.At present,the HIP platform provides access to multiple European centres and hosts Intracerebral EEG(IEEG)patient datasets.IEEG data is a valuable and limited resource,and its integration has the potential to drive significant progress in the fields of epil

249、epsy and brain research(20).A third and even more recent endeavour is the EBRAINS HealthDataCloud(EHDC).The EHDC is being developed by a mixture of existing Human Brain Project(HBP)/EBRAINS infrastructure partners and leading health data service providers that have recently joined the HBP.The Health

250、DataCloud aims to provide EBRAINS services for sensitive data.The (19)https:/www.ebrains.eu/tools/medical-informatics-platform (20)https:/www.ebrains.eu/tools/human-intracerebral-eeg-platform Human Brain Project-10 Years Assessment 22 EBRAINS HealthDataCloud is a federated research data ecosystem of

251、 interoperable nodes including a central node deployed at EBRAINS RI and an expandable set of satellite nodes deployed at hospitals,research institutes and computing centres(21).The EHDC seems to include a mixture of a federation of nodes with a central server and has a much larger and broader scope

252、 on brain health,promising an approach for sharing data that is potentially powerful,provided security and GDPR compliance remain safeguarded.Services provided by these EBRAINS platforms have the potential to bring benefits for society,by enhancing neuroscience research and,in this way indirectly in

253、 the medium to long term,to an improvement of healthcare services related to brain diseases.A valuable contribution towards the enhancement of the research and innovation capacity of the partners involved has been provided so far.The future will learn how the full usability and uptake of these platf

254、orms by the community is ensured,especially by the medical community at large.Finally,the new EBRAINS 2.0 project is building on the efforts of the HBP in the field of Medical Neuroscience,and developing a network of European hospitals,together with the European Academy for Neurology,to gather uniqu

255、e and high-quality neuroimaging data of patients.4.6.Simulation Neuroscience HBP researchers were among the first to understand that,in order to develop accurate and comprehensive brain simulations,it is fundamental to investigate the dynamics and organization of brain networks across these diverse

256、scales(DAngelo&Jirsa,2022).In the original HBP proposal,the Brain Simulation Platform had the goal of integrating data in multi-level models,starting from the mouse up to the human brain.By utilizing supercomputing technology,the platform aimed to simulate and visualize the models behaviour.Validati

257、on against biological experiment results would have identified data gaps,ensuring continuous improvement in model accuracy throughout the project.Now,after a decade of dedicated effort,HBP researchers have advanced brain simulation by simultaneously exploring multiple scales.This breakthrough enable

258、s them to study the brain in ways never before possible.The brain simulation activities have been probably one of the most successful of the entire project,both in terms of results,dissemination,and impact.Many of the developments started as the activities of a few partners and focussed on so many d

259、ifferent problems that an integrative view was very difficult to appreciate.Notwithstanding,thanks to the significant effort done within the last phase of the HBP,most of modelling results are now part of EBRAINS,the (21)https:/www.healthdatacloud.eu/Highlights,Outcomes and Impact Significant progre

260、ss in brain simulation by integrating data across scales,thus revolutionizing our understanding and predictive capability of brain functions Personalised modeling,especially in epilepsy treatment and clinical trials High-resolution data and modeling techniques to drive modern neuroscience towards pe

261、rsonalized diagnostics and therapy,marking a significant advancement in medical approaches Human Brain Project-10 Years Assessment 23 HBP centralized platform intended to accommodate all the diverse models and results and now available for the entire scientific community(Schirner et al.,2023).Over t

262、he years,brain simulation methods developed in the HBP have expanded in both range and predictive capability.Starting from the early development of The Virtual Brain(TVB)simulation engine(Ritter et al.,2013;Woodman et al.,2014;Sanz-Leon et al.,2015)al.,2014),a lot of improvements have been made in e

263、pilepsy treatment,where the prediction of seizures and the outcomes of surgical procedures can now be achieved through brain models personalized to the individual patient data(Jirsa et al.,2017;Makhalova et al.,2022)2).Thus,thanks to the HBP,a novel connection between brain simulation and clinical p

264、ractice has begun to be possible,with personalised modelling emerging as an unforeseen but pivotal advancement in computational modelling and simulation.Significant strides have been made,notably with the integration of digital twin technology into medical practice thanks to HBPs implementation of p

265、ersonalized modelling through TVB.This innovative approach is currently undergoing testing in the EPINOV clinical trial for epilepsy,running in France and involving 13 hospitals(Wang et al.,2023).By capitalising on those results,HBP researchers have started utilizing multi-scale simulations to study

266、 other pathologies than epilepsy(Jirsa et al.,2023),such as Parkinsons disease(Meier et al.,2022),where simulation of the DBS therapy could help clinicians further optimize the treatment.Importantly,as a platform for whole-brain modelling,TVB is now integrated with high-resolution atlas data and hig

267、h-resolution neural network modelling using NEST in Co-Simulation frameworks(Goldman et al.,2023;van Keulen et al.,2023).Many other key achievements are worth mentioning:experimental and computational studies to study how the morphological properties and dendritic structure of neurons influence thei

268、r behaviour and impact phenomena at higher levels of brain organization,including consciousness(Goriounova et al.,2018;Gidon et al.,2020);investigating the role of neuronal networks in perception(Filipchuk et al.,2022);unravelling the structure and function dynamics in cortical microcircuits of the

269、cerebellum(De Schepper et al.,2022);application of modelling techniques in HBP to better understand and/or classify the evolution of different pathologies or physiological states such as ALS(Polverino et al.,2022),Multiple Sclerosis(Sorrentino et al.,2022),Alzheimers(Triebkorn et al.,2022),treatment

270、-resistant depression(An et al.,2022),and brain ageing(Escrichs et al.,2022).The HBP has facilitated the progress of researchers in enhancing diverse computational brain modelling and simulation strategies,partly through their integration with EBRAINS,pioneering the development of innovative multi-s

271、cale models.The links between anatomical and functional data using multiscale modelling at the cellular level seem to show converging benefits that deserve further development.Such modelling and simulation strategies will not only enhance research outcomes but also serve as a fundamental principle d

272、riving modern neuroscience,where Digital Twins of the individual brain will serve to improve personalized diagnostics and therapy(Amunts et al.,2022).Thanks to these results,the HBP has pioneered a truly personalised medical approach.Human Brain Project-10 Years Assessment 24 4.7.Neuro-inspired AI T

273、he lack of mention of AI(Artificial Intelligence)in the original proposal of the HBP signifies the evolving nature of scientific developments in the field over the last ten years.Initially focused on computational neuroscience,the project later incorporated AI with the goal of exploiting neuro-inspi

274、red learning mechanisms to improve machine-based methodologies and techniques.This progression and evolution highlight the pivotal role of neuroscience as a means of translating brain research into tangible outcomes capable of driving the advancement of AI(Verzelli et al.,2024).Indeed,when the HBP w

275、as proposed as an EU FET Flagship back in 2012,it demonstrated a strong interdisciplinary nature,where boundaries between different fields blurred and new approaches could emerge.It is worth highlighting that two of the HBPs initial goals were(i)Theory to identify mathematical principles underlying

276、the relationships between different levels of brain organization and their role in the brains ability to acquire,represent,and store information;(ii)ICT platforms as an integrated system offering services to neuroscientists,clinical researchers,and technology developers that accelerate the pace of t

277、heir research.Given the above,the project clearly was in the perfect position to advance AI-based approaches,thanks to its strong interdisciplinary nature and focus on both computational modelling and ICT.From its inception,the HBP put a lot of effort into studying the brains learning rules at diffe

278、rent scales to be further exploited for machine-based applications(Amunts et al.,2024).All neuronal/neural networks,whether found in living organisms or created artificially,rely on plasticity processes for learning.While neuroscientists have collected a wealth of data on plasticity,even within the

279、HBP,a comprehensive theoretical framework to tie it all together was lacking.On the other hand,deep learning techniques in artificial neural networks could be systematically developed,but they often involve elements that do not mirror biological processes.In the last phase,HBP researchers worked tow

280、ards bridging this gap.The work aimed at addressing the intricacies of brain complexity at both temporal and spatial scales,to unravel the influence of various properties of brain networks(e.g.,plasticity,topology,modularity)on resultant behaviour(Schirner et al.,2023).Indeed,most of the activities

281、involved the development of brain-inspired models and learning rules to be more biologically plausible,with the final aim of advancing innovative applications in robotics,automation,and AI.In this regard,the modelling approach combined brain-inspired architectural principles with parameters derived

282、from biological learning.Modular design allowed a single module to serve multiple architectures.Outcome architectures were expanded to include higher-level cognitive functions and embodied in robots for tasks like navigation and manipulation.Some of the key achievements of these activities include:b

283、iologically plausible hierarchical memory models such as the BrainProp algorithm for attentional feedback(Pozzi et al.,2020);the development of recurrent neural networks trained with a biologically plausible learning rule to solve multistep visual routine tasks(Kroner et al.,2020);hierarchical visua

284、l processing(Kroner et Highlights,Outcomes and Impact Exploitation of neuroscientific results to advance machine-based methodologies for AI Interdisciplinary efforts in studying brain learning rules to develop more biologically plausible models Development of biologically plausible models to enhance

285、 robotics,automation,and AI applications Human Brain Project-10 Years Assessment 25 al.,2020);visual relational reasoning(Thompson et al.,2022);models for spatial planning and navigation(Muhle-Karbe et al.,2023;Stckl et al.,2024);compositional inference,i.e.,“assemble”knowledge from different source

286、s(Nelli et al.,2023;Max et al.,2024).In parallel,HBP researchers investigated the impact of attention,synaptic tags,feedback connections,and neuromodulators on learning.Their focus was on combining the development of biologically plausible learning rules within neural networks with the ability to le

287、arn and adapt similarly to the human brain.Some of the main outcomes encompass:the biologically plausible implementation of natural-gradient-based plasticity for spiking neurons(Kreutzer et al.,2020);the investigation of how task-dependent top-down signals can reshape the functional mapping of senso

288、ry processing networks at the micro and mesoscopic levels(Wybo et al.,2023);the emergence of predictive coding through synaptic plasticity in hierarchical cortical areas(Dora et al.,2021).It is important to notice that,due to the highly inter-and multi-disciplinarity of the topics entailed in these

289、activities,interactions among partners were often significantly more challenging than in other activities.This led to the formation of small sub-groups of institutions and scientists,each undertaking different experiments and parallel studies simultaneously.Indeed,HBP activities in this area appears

290、 quite fragmented,without a clear coherence and a small direct impact on the broad range of potential applications foreseen in the original proposal(cf.Neuromorphic and Neurorobotics sections).As of today,few of the developments have been integrated in the EBRAINS platform or tested in relevant envi

291、ronments.Nevertheless,the results obtained achieved relevance among both neuroscientists and computer scientists,and many of them were published in high-impact journals,denoting the scientific quality and impact of the research performed.4.8.Robotics Modern AI developments,such as large language mod

292、els and image recognition systems,can nowadays replicate cognitive functions previously thought to be exclusive to biological brains.However,these models lack bodily experience and environmental interaction.Similarly,typical computer models of the brain used by neuroscientists,even if complex and us

293、eful,may not fully capture the richness of real-life cognition,which involves the interaction of mind,brain,body,and environment.Indeed,behaviours,ranging from the simplest reflexive actions to the most intricate cognitive processes,are shaped through the continuous interplay between an organism and

294、 its environment.The brain orchestrates these behaviours Highlights,Outcomes and Impact Pioneering advancement in embodied cognition by connecting brain models with robotic bodies,promoting understanding of cognition through embodied experiments.Neuro Robotic Platform(NRP)-based innovative experimen

295、ts,such as multisensory integration and precise movements,fostering research in neuro-derived AI and robotics.A versatile framework for integrating functional models,enhancing research in computational neuroscience and embodied AI,and paving the way for complex cognitive architectures and emergent p

296、henomena studies.Human Brain Project-10 Years Assessment 26 through the coordinated activation of neuronal assemblies,dynamically adapted depending on the task.To the HBP,it was clear that to emulate brains and behaviours,embodiment was not only necessary but fundamental(Prescott&Wilson,2023).To thi

297、s end,the HBP community has pioneered new scientific tools that allow connecting brain models with robotic bodies to study processes of embodied cognition thanks to a framework named Neuro Robotic Platform(NRP),designed from the early phases of the HBP.The goal of NRP was indeed to provide an experi

298、mental setting to test the capabilities of the developed brain-derived models,to be further exploited also in the industrial context.The idea was to provide the EU research community with robots equipped with human-like brains,capable of better cognition(Schirner et al.,2023).To this end,the HBP sci

299、entists started designing several neurorobotic solutions,allowing them to perform fully simulated experiments(virtual brain and virtual robot)and hybrid experiments(either a physical or virtual robot connected to a simulated or neuromorphic brain)in a fully closed-loop fashion.Some of the key achiev

300、ements include:multisensory information integration,such as the MultiPredNet based on predictive coding and able to integrate visual and tactile information(Pearson et al.,2021);object manipulation with high degree of dexterity,provided by AngoraPy(Anthropomorphic Goal-Driven Responsive Agents in Py

301、thon)which enables neuroscientists to build and train neural network models with a brain-derived architecture in ecologically valid settings using reinforcement learning and goal-driven modelling(Weidler et al.,2023);precise movements and coordination through a neuro-derived AI system(e.g.,interacti

302、on between an artificial neural network mimicking the cerebellum and a robotic arm(Abada et al.,2021;Antonietti et al.,2022);cobotics(i.e.,robots designed for direct interaction with humans in a shared space)simulations developed using EBRAINS in which robots learn to interact with humans in a safer

303、 way,to allow humans to better trust collaborating with robots well demonstrated in Showcase 6(Feldotto et al.,2022;Stolpe&Morel,2023);integration of The Virtual Brain large-scale as well as NEST simulations into the NRP;generation of synthetic data as input for Machine Learning experiments;integrat

304、ion with major HBP neuromorphic technologies such as SpiNNAker and BrainScaleS.Thanks to its modular architecture,open-source nature,and comprehensive online documentation,the NRP offers computational neuroscience and embodied AI communities a versatile framework for seamlessly integrating multiple

305、functional models.Notably,the NRP enables each model to operate and communicate at different frequencies within the simulation time domain,mirroring the diverse neuronal activities observed in the brains anatomical areas and thus providing robots with brain-derived skills.This unique capability woul

306、d support the future development of complex cognitive architectures with diverse components,facilitating the in-depth study of emergent phenomena.Furthermore,the NRPs use of container technology ensures its adaptability and readiness for integrating new simulation software,AI systems,or brain models

307、 without concerns about dependency conflicts.Despite these results,however,the integration of the activities of the NRP with the outcomes from activities related to brain modelling(i.e.The human multiscale brain connectome and its variability)and neuroscience(i.e.Networks underlying brain cognition

308、and consciousness)has been only partially addressed by the consortium.The robotic application has been around since the projects beginning(cf.HBP proposal),so a focus on more integrative work was the aim during the final years.Notwithstanding this aim,a strategic view of the research as a whole,and

309、the outcomes across the possible(many)robotic applications,was not successfully addressed.At the end of the project only a few examples(e.g.,Showcases 5 and 6)were provided.Furthermore,the relationships of the activities with respect to EBRAINS are still not clear,thus limiting the impact of the pla

310、tform for real-world applications.Human Brain Project-10 Years Assessment 27 4.9.Integration platform for digital neuroscience(EBRAINS)The research ecosystem developed by the Human Brain Project,subsequently referred to as EBRAINS,is a crucial component of the digital neuroscience field.With approxi

311、mately 160 tools integrated,and an average Technology Readiness Level(TRL)of 7,the EBRAINS ecosystem provides a robust framework for advancing brain research.The successful integration of a diverse portfolio of tools alone constitutes an important milestone for the advancement of digital neuroscienc

312、e.The adoption of the platform has successfully advanced through the HBP.By September 2023,the EBRAINS community boasted 8475 users from around 1500 institutions,fostering a collaborative environment for further exploration of the brain.This number has increased after the end of the HBP to more than

313、 10.000.Among its facilities EBRAINS includes the Knowledge Graph,an impressive metadata management system that enables data annotation and curation.This ensures effective management of the data,software and models integrated on the platform,which can be navigated interactively.EBRAINS currently inc

314、ludes the following(22):Comprehensive human,macaque,and rat brain atlases;Numerous datasets for experimentation and analysis;Modelling,Simulation and Computing Platform.Validation and Inference Services;Health Research Platforms.The capability of working with sensitive data is mainly achieved throug

315、h the delivered Health Research Platforms.Here the HBP has successfully paid special attention to achieve GDPR-compliance and cover all the regulatory requirements for handling sensitive data.It is worth mentioning that the“Human Data Gateway”service(inside the Sensitive Data stack of EBRAINS Servic

316、es)is assumed by the Health Data Cloud(HDC)platform,which is part of (22)https:/www.ebrains.eu/Highlights,Outcomes and Impact EBRAINS integrates 160 interoperable tools facilitating its access to a community of approximately 8500 users from around 1500 institutions around the world(as of Sept 2023).

317、Initially developed at HBP,the integrated High Performance Computing infrastructure is being further developed in the ongoing ICEI project.Potential health applications are exploited in sequel projects eBRAIN-Health,TEF-Health and AI-MIND.FAIR data services,and open software facilities crystallize t

318、he platforms commitment to open science encouraging the development of complex collaborative breakthrough approaches for digital neuroscience.Human Brain Project-10 Years Assessment 28 Health Research Platforms together with the MIP and the HIP(see Medical Neuroscience section above).Modelling servi

319、ces offer a variety of simulation software tools for multi-level brain modelling.These tools are enabling the developed virtual models of neurological conditions like epilepsy,dementia,and cognitive processes associated with perception and memory.They provide integrated access to the data required f

320、or implementing model workflows of unprecedent complexity(see Showcases section).Overall,the EBRAINS platform provides researchers with a comprehensive ecosystem for conducting the most advanced brain research,with robust data and model services that support the construction of intricate workflows.T

321、he EBRAINS platform integrates a range of modern digital toolkits and libraries crucial for modelling and simulation in digital neuroscience.These tools are meticulously designed to address various aspects and scales of brain function and structure,empowering researchers to conduct detailed and grou

322、nd-breaking studies.Among the comprehensive collection of 160 tools,some are particularly noteworthy due to their pivotal importance and strategic position within the EBRAINS ecosystem:1.Knowledge Graph:a metadata management system developed to find and share software,models and data.It includes fun

323、ctions for search,editing,query building,statistics and APIs,and manages 1064 data sets,4 metadata models,225 pieces of software and 18 webservices from 2195 contributors.2.OpenMINDS:The Open Metadata Initiative for Neuroscience Data Structures is a community-driven metadata framework for neuroscien

324、ce graph databases that was empowered by HBP and EBRAINS.The openMINDS metadata models are adopted by the EBRAINS Knowledge Graph,the EBRAINS Atlas Service,and The Virtual Brain(TVB).It is currently in the process of being adopted by the Japan Brain/MINDS project.(23)3.EBRAINS Curation Workflow:The

325、curation workflow ensures that metadata is accurately entered into the Knowledge Graph according to the OpenMINDS framework.It organizes data properly and makes it accessible in a standardized and interoperable format with a Data Descriptor.Additionally,it links data to relevant analytical tools.4.S

326、iibra toolsuite:this software has been developed for interacting with brain atlases and is including a siibra-explorer,an interactive 3D atlas viewer and a dedicated programmatic Python client.It is tightly integrated with the EBRAINS Knowledge Graph(24),allowing the seamless querying of semanticall

327、y and spatially anchored datasets for the exploration of the regions of the human,rat,mouse and monkey brains at microscopic detail,as well as the discovery of related multimodal data features.siibra-python is offering an easy and well-structured way to include maps,reference templates,region defini

328、tions and linked datasets into reproducible programmatic workflows.(23)https:/brainminds.jp/en/(24)https:/kg.ebrains.eu/Human Brain Project-10 Years Assessment 29 5.The Virtual Brain(TVB):TVB(Ritter et al.,2013)is a crucial component within the EBRAINS platform serving as a powerful tool for whole-b

329、rain simulation.Its purpose is to integrate empirical data into personalized brain network models,offering a comprehensive approach to understanding brain function.TVB is specifically designed to construct,run,and integrate neural mass models,enabling researchers to study the dynamics of large neuro

330、n populations.The TVB is emerging as being potentially pivotal in neurological studies,including those on brain ageing,disease progression in neurological disorders,and the effects of neurosurgical interventions.A strong indication of its success is the integration of the TVB into several public-fun

331、ded projects,highlighting its significance within the scientific community and the suitability to unleash the potential of digital neuroscience already today.6.NEST:The purpose of NEST(Gewaltig&Diesmann,2007)is to facilitate the simulation of large-scale neuronal networks,with a particular emphasis

332、on spiking neurons,in order to better understand the dynamics of the brain.NEST offers efficient and scalable solutions that can be implemented in high-performance computing(HPC)environments,making it an ideal tool for conducting large and complex simulations.Its ability to run simulations on HPC fa

333、cilities has been recognized as one of its most notable features(Tikidji-Hamburyan et al.,2017).NEST is commonly used in research studies that investigate brain connectivity,learning processes,and the emergence of neural patterns.It serves as a valuable resource for studying and analysing these neuroscientific phenomena.7.Neuron:NEURON(Carnevale,2007)is used for biophysically detailed neuronal mod