ACC
ACC is contributing to the development of HBD with several projects, 6 of which are loaded into the
Phases
Phase One
The first step of the project (2020-2023) was concluded with the achievement of three objectives.
Objective 1. Creation of a first draft of an Integrated and Federated National Platform within INFN (Cloud Platform INFN-HBD).
The INFN-HBD Cloud is a Community Cloud based entirely on open source components, inspired by the NIST Cloud Federation Reference Architecture (federated model) and designed to ensure interoperability and portability of data; calculation capacity, network throughput and storage; availability of services and data; protection of personal information, regulatory compliance (NIS, NIS2, GDPR) and multi-tenancy (isolated virtual environments); security (confidentiality, integrity, availability and privacy): SGSI, ISO/IEC, ENISA); implementation of evolved governance models and policies; reversibility of data sharing. Cloud federation is achieved through four key systems: authentication and authorisation; data management; data transfer; automatic resource selection based on access policies, requests and resource availability.
Objective 2. Migration to the INFN-HBD Federated Cloud Platform of the retrospective data generated by the Network Projects.
The transfer to the Cloud of retrospective data generated by specific Network projects (data owned by the Networks) represented a first level of validation of the INFN-HBD Federated Cloud Platform, with regard to the type of data ingested and the federability of other data collection platforms.
In particular:
i) ACC Network: the import of genomic and radiomic data from the MELANOMA, ACC-Lung, and ACC Rationale Network Projects, ACC RECORD project, GERSOM PROJECT, for a total of about 6 TB of space, was successfully started.
ii) RIN Network: the resources (about 30 TB) and protocols for importing MRI data from the Redcap safe, XNAT- CLUSTER, RIN-Imaging, Neuro RIN projects and related genomic data (where available) were prepared.
iii) Cardiology Network: the federation process with the Consortium for Bioengineering and Medical Informatics - CBIM, which manages the RedCap platform of the Cardiology Network (Projects VavirimS, Improve, SKIM, SPB, Outstanding, Fate, CophyTea, CardioCovid, Carditox, CV Prevital) was successfully started.
Objective 3. Mapping the level of data management expertise of each IRCCS and sharing of data extraction/management protocols (interoperability).
Protocols for the extraction/management of structured (omics, signals and imaging) and unstructured (clinical texts; EMR; CTMS) data, templates and policies for managing data privacy, security and ethics, and some pipelines for advanced analysis of clinical and imaging data were generated and shared with participating IRCCSs. This was achieved through the activity of 7 Working Groups (WGs), each dedicated to a specific topic, composed of IRCCS representatives and coordinated by experts from the Politecnico di Milano.
WG1. NLP (Natural Language Processing).
WG2. Omics, imaging and signal data.
WG3. Interaction between electronic health records and clinical data management systems for research purposes.
WG4. Identification of the methodological and technological tools required for data extraction and integration.
WG5. Physical system architecture.
WG6. Data Analysis.
WG7. Identification and implementation of solutions related to privacy and security, and other ethical issues.
Phase Two
The second step of the HBD Project (2024-2026) will focus on three main objectives.
Objective 1. Validation of the INFN-HBD Federated Cloud Platform.
This validation is based on the generation and sharing of prospective data generated by the participating IRCCSs. A number of multi-centric scientific projects will be implemented involving the participating IRCCSs through the related Networks (Special Projects), and involving: i) the generation of clinical-scientific data at the IRCCSs according to the data processing protocols established and validated in the first phase; ii) their ingestion into the INFN-HBD Cloud Platform; and iii) their sharing by the related IRCCSs.
Objective 2. Extension of the INFN-HBD Cloud Platform to other biomedical data collection projects.
Over the past three years, a series of contacts have been established to explore synergies of the HBD Project with other biomedical data-science projects, from which a widely shared system vision has emerged. The objective of the next three years is to formalise these contacts in order to identify and implement specific projects and suitable institutional frameworks for the federation and systemisation of the most advanced biomedical data-science initiatives in the country, including the possibility of using INFN-HBD as the national cloud of clinical-scientific research data.
Objective 3. Development of advanced analytics algorithms (artificial intelligence- AI) and creation of a community of scientists dedicated to the analysis of biomedical big data (Project HBD-AI).
The ultimate goal of the HBD project is to create a universal knowledge resource in healthcare, based on free access, by all the various stakeholders, to real-time data from hundreds of thousands of patients. The use of artificial intelligence represents a potentially transformative analytical approach. Crucial to the full realisation of AI’s potential in healthcare is the development of more accurate and reliable AI algorithms, the creation of a community of data-scientists dedicated to the analysis of biomedical big-data, and the guarantee that AI is used safely and fairly.
The central role of Networks
The four Networks play a central role in the HBD project: each, in the start-up phase, identified an already active research project and shared it on the platform.
This has enabled HBD to work quickly with existing and heterogeneous data assets: omics, clinical practice, imaging or, in anticipation, directly collected from patients and/or sensors.
Currently, the networks have identified and shared with HBD further lines of research that have data standardisation and subsequent sharing, also with the IRCCSs participating in HBD, as their common denominator.
RIN
The RIN Network has different platforms on which research study data are managed.
Cardio Network
The Cardio Network already has a shared dataset available for individual research projects.
IDEA Network
The IDEA Network is actively contributing to HBD with an important pilot project.