The Data Provider Layer is a virtual layer that connects third-party software with the Federated Data Management and Processing layer. It allows the processing layer to access real-time data from various sources, such as EHRs with FHIR standards, administrative datasets for socio-economic status, and Copernicus S-5p mission. It uses custom connectors (e.g., JDBC) to retrieve structured and unstructured data and metadata from the healthcare facilities, the public administrations, and the regional information systems.

The Federated Data Management and Processing Layer takes data from the Data Provider Layer and performs operations such as manipulation, integration, and aggregation. This layer is the computational core of the architecture and it uses federated learning to combine the results of machine/deep learning models trained in different healthcare facilities. It balances and optimizes the methods to achieve a robust and reliable risk assessment system. It also handles local access, pseudonymization, data quality assessment, training-test-validation phases of distributed machine learning models, and GDPR compliance. It uses domain ontologies to harmonize the data syntactically and semantically, representing the global model of the data space and enabling access to it.

The Big Data Analytics and Data Visualization Layer receives the output of ML/AI processing and provides statistical analysis, dashboarding, data visualization and decision support services to the end-user. It has an analytics module that enables data exploration and system performance monitoring through appropriate metrics and KPIs. It also implements and exposes all business intelligence and data visualization functionalities. This layer includes the “Security Control” module is a cloud-based module that ensures the security of the entire infrastructure. It monitors cloud configurations, detects suspicious activity, prevents insecure deployments and limits excessive privileges for Identity and Access Management (IAM) roles.