THE HUMAN BRAIN PROJECT FLAGSHIP 1 ST TECHNICAL PROJECT REVIEW (PERIOD COVERED FROM 01/10/2013 TO 30/09/2014) MAIN REVIEW CONCLUSIONS AND RECOMMENDATIONS INTRODUCTION The reviewers acknowledge the good quality of the work that has been carried out by the consortium in the first year of the Human Brain Project (HBP) and recognise the huge task the creation of the HBP flagship represents but recommend also the implementation of important corrective actions. In the majority of the subprojects, work packages and tasks, good progress has been made towards the objectives, as defined in the HBP Description of Work (DoW). Many of the individual subprojects are generating important new data and there has been substantial progress in the development of the ICT platforms. In this respect the project has benefited from being to a large extent the continuation of more than a decade of previous, wellsupported, high quality research projects, e.g. the Blue Brain project, the FACETS and BrainScales projects, and the SpiNNaker project. The leaders of the project and of the subprojects have demonstrated that they maintain a clear vision for the HBP and that their firm ambition is to achieve the overall goal of HBP. However, whilst the participants were all excited by and fully engaged in their own subprojects, it is clear that very significant efforts remain to be made, in terms of coordination and integration, for the HBP to become a truly large unified project. There is a clear need for a tighter and more carefully managed integration and realignment of the work in the Data and Theory subprojects, both with the development of the ICT platforms, and within and between these subprojects themselves. A more rigorous methodology for infrastructure construction and operation is also required for ensuring success in translating the platforms into a solid ICT integrated infrastructure. Moreover, it is crucial that the consortium engages with the wide scientific community in the co-design and development of the ICT platforms. In the first year, the consortium has set up all the governance and management structures as proposed in the DoW. However, the reviewers recommend that changes are made to ensure that the decision making processes are simple, fair and transparent. The Consortium has also initiated an effective dissemination mechanism to promote the project to the general public, including an ambitious science museum initiative, resulting in a substantial interest worldwide. The consortium has also developed a solid education programme (HBP Curriculum, education workshops, HBP schools) and has initiated a number of international scientific contacts with other related projects and institutions, in the USA, Japan, China and Australia, which are expected to lead to extensive and productive exchanges of data and expertise. It is important for the HBP to better articulate its strategic goals and to communicate them in a clear and realistic way, within the HBP, to the wider scientific community and to the public, and to avoid at all costs creating unrealistic expectations. The goals must be communicated to 1
the scientists outside HBP in a manner which allows an engagement in open debates that would help clarify the scientifically and technically achievable targets of the project. IMPLEMENTATION OF CORRECTIVE ACTIONS 1. Closer integration of the Data and Theory subprojects with the development of the ICT platforms The HBP is an exceptionally ambitious and highly interdisciplinary project, formulated as a very large integrated flagship research initiative. It is structured into thirteen subprojects: four Data and Theory subprojects (SP1 to SP4), six ICT platform subprojects (SP5 to SP10) and three cross-cutting subprojects for Applications, Ethics and Society, and Management (SP11 to SP13). The Data and Theory SPs are described as follows in the DoW: Data (SPs 1-3). This work will collect the neuroscience data required for the operation of the ICT platforms. The data collected during the ramp-up phase will feed the models and simulations developed during the operational phase. Theory (SP 4). This work will develop the theoretical frameworks necessary to link brain models and simulations representing different levels of brain organisation. So far, the HBP is not developing with the expected level of integration and the project controls in place are not adequate to achieve this aim. There is a clear need for a tighter and more carefully managed integration and realignment of the work in the Data and Theory subprojects, both with the development of the ICT platforms, and within and between these subprojects themselves The onus of responsibility in achieving the necessary integration and alignment is not solely that of the participants in the Data and Theory subprojects. It is vital that the ICT platform developers clearly articulate their needs and that they also look beyond the HBP for the necessary data and theory to achieve their objectives, ideally through collaboration with partnering projects, but also generally from the international scientific community. The Data and Theory SPs (SP1 to SP4) must be the initial users of the ICT platforms and be deeply engaged in their co-design. The consortium should develop and put into place, by June 2015, a detailed strategy and plan to effectively integrate and align the work in SP1 to SP4 with the platform developments within the project. This should include plans to better align Data producing activities in SP1, SP2 and SP3 with the major aim of the HBP to ensure that SP1 and SP2 results feed into multi-level models of the mouse and human brain, and that the SP3 results guide the development of high-level models of neuronal circuitry. The coordination between SP1, SP2 and SP3 needs a clear definition of the questions to be asked. This requires urgent strategic decisions about mouse strains to be used, the types of data required for the modelling goals, the best techniques to obtain those and the links between mouse and human work. Regarding SP4 the plans must align the theory activities with the need to support the modelling work in SP6 and SP9 in particular. As part of this strategy and plan the consortium should also develop and put into place, 2
as soon as possible, a strong mechanism for collaboration between the subprojects SP1 to 4, and establish a project Task Force having as its role: (i) to align the Data and Theory activities (this includes, for example, designing new experiments for data acquisition in SP1 and SP2, and planning the acquisition of new strategic data from outside the project, as needed); and (ii) to continuously monitor the inter-collaboration of the four Data and Theory SPs and their interaction with all the other SPs, in particular SP5 and SP6. The Task Force members should include scientists from both inside and outside the project. As announced by the HBP at the review meeting around 10% of its budget should continue to be spent on studying cognitive architectures in future budget allocations. In order to ensure that the HBP takes full account of ongoing international scientific developments in addition to those within SP1 to SP4, the consortium should ensure the full dissemination of the results from these SPs, and the initiation of an active debate with the international scientific community on the Data and Theory required for developing the platforms. The European Institute for Theoretical Neuroscience (EITN) should act as a major forum for stimulating such scientific debates, thus allowing the project to validate or reconsider its orientations and developments in respect of the project's next planning and funding phases. 2. Achieving the goal of an integrated ICT infrastructure for the scientific community The HBP is developing six ICT platforms, dedicated respectively to Neuroinformatics, Brain Simulation, High Performance Computing, Medical Informatics, Neuromorphic Computing and Neurorobotics. These platforms will be made accessible over the Internet via an HBP Unified Portal to the scientific community in neuroscience, medicine and computing. Achieving this goal is entirely dependent on providing a high quality, integrated ICT infrastructure which is adopted by the scientific community as a major and central tool to support their work. There are two inter-related issues here, that of ensuring the high quality and integrated nature of the ICT platforms and that of ensuring that the infrastructure meets the needs of the community to the extent where it becomes a de facto tool in their work. Integrated infrastructure development It is essential that the HBP management ensures a high level of quality control over the development and integration of the ICT platforms. To this end, a more rigorous infrastructure building and operating approach is needed for ensuring success in translating the platforms into a solid ICT integrated infrastructure. A specific management structure has to be put in place in order to ensure the integration of all the platforms into such an infrastructure, including data management, quality assurance, interfacing and software engineering. The consortium must provide by June 2015: plans for system engineering covering (inter-)platform integration specifications, testing, verification and validation and phased deployment; a quality and technical risk management plan for the data and software modules; and the specifications for integrating modules and data; 3
explicit specifications of the data infrastructure (data types, metadata, vocabularies, interfaces); agreed rules for handling IPRs and copyright. Building an HBP user community The consortium must immediately start to address the need to build a user community for the ICT platforms, and to engage both the HBP partners and the wide scientific community in the co-design and development of the platforms to meet their needs. This should start, as planned, with the release of the platforms within the HBP community, but should extend to a set of international early-adopters outside of the HBP at the earliest opportunity. The required actions include: developing a concrete plan for attracting and involving as many and as wide a range of users of the platforms as possible, as quickly as possible, and including identifying concrete incentives that the HBP offers in order to engage them, and providing high quality and continuously updated user training and documentation; defining metrics of success in engaging users and closely monitoring these; releasing as many platform components as soon as possible, firstly within the HBP and as soon as possible thereafter to the wider scientific community, and proactively disseminate them (a good example of such components is the NEURON simulation component of the Brain Simulation platform, and a good starting point for dissemination and user recruitment is FENS); specifically for SP8 (medical informatics platform), defining a proactive approach for engaging with hospitals, and integrating as soon as possible a number of different hospitals, starting with one hospital in Germany and one in the UK. 3. Effective organisation and management of the project There are several issues in the organisation and management of the project which need addressing as detailed below: Governance and administration The governance and decision making processes need to be changed to ensure that decision making processes are simple, fair and transparent. In order to be in place for the operational phase, some changes to the management structure and its functioning must be already implemented in the ramp-up phase and should be communicated without delay to the Commission and to the HBP partners. Interaction and communication between SPs should be substantially improved. The Consortium should consider creating two or three clusters that will group a subset of SPs to ensure closer communication and better interactions between these SPs. In particular, SP1-4 should be grouped in a cluster and the platform SPs grouped in another cluster. The interactions of SP12 with all SPs have to be strengthening, and a clear plan for ethical management should be provided. 4
There is a need to simplify and streamline project administrative activities and adapt accordingly the number of people involved in running such activities; this includes reconsidering internal project reporting (to six months rather than quarterly), scientific Key Performance Indicators (KPIs), project-level KPIs, etc. Reporting Project deliverables should be concise and to the point, and, if needed, refer to internal project reports serving the communication and interworking needs of the consortium. The Periodic Progress Report (PPR) must fully, clearly, concisely and accurately report the scientific progress and results achieved, according to the subprojects, work packages and tasks of the DoW. It should complement the deliverables in order to allow a straightforward assessment of the work done, by including a critical selfassessment, i.e. the scientific or technical merit of the results achieved and the difficulties encountered, if any. The deliverables and the scientific results sections of the PPR must remain under the responsibility of the scientists and developers in charge of the work. Communication It is necessary for HBP to better articulate its strategic goals and to communicate them in a realistic way, within the HBP, to the wide scientific community and to the public. These goals and how they will be achieved must be formulated by the HBP as a whole, with the Coordinator and the SP leaders working in close collaboration with the Work Package and Task leaders to ensure that they are clear and realistic. The communication of the goals must also be clear and avoid at all costs creating unrealistic expectations. The goals must be shared with all the researchers inside HBP so that they are fully aware of their personal role in the project in contributing to the goals. The goals must be communicated to the scientists outside HBP in a manner which allows an engagement in open debates that would help clarify the scientifically and technically achievable targets of the project. SP12 (ethics and society) should assist in the communication and engagement with the public in coordination with SP13 (management). 5