The FAIR principles provide a framework to ensure that scientific data is accessible, reusable and properly managed. Their implementation is a central objective of the European Open Science Cloud (EOSC).
However, enabling access to sensitive or confidential data while preserving privacy and maintaining its usability for research remains a major challenge. Traditional solutions such as safe rooms or restricted environments often limit accessibility and hinder reproducibility, creating tensions with the goals of open science and FAIR data.
EOSC-SIESTA (Secure Interactive Environments for SensiTive data Analytics) addresses this challenge by developing a set of tools, services and methodologies that enable the secure sharing and analysis of sensitive data within EOSC. The project follows a cloud-based approach, providing trusted environments where data can be accessed and processed without being exposed.
SIESTA focuses on delivering user-friendly tools that facilitate the adoption of privacy-preserving workflows. These include secure infrastructures and advanced anonymisation techniques, designed to support data sharing while protecting confidentiality.
To support users in adopting these solutions, EOSC-SIESTA provides tutorials and webinars where researchers and practitioners can learn how to effectively use the project’s tools and services. These activities aim to lower the barrier to entry and promote the practical implementation of privacy-preserving data analysis workflows.
The overall objective is to enhance EOSC Exchange services by enabling the analysis of sensitive data in trusted cloud environments, demonstrating that FAIR principles can be effectively applied even in scenarios involving restricted data.
The SIESTA project has contributed to the scientific community through peer-reviewed publications, technical deliverables, software and data releases. You can explore the complete list of outputs in our Zenodo community.
The main publications concerning peer-reviewed journal and conference papers include:
- Sáinz-Pardo Díaz, J., Athanasiou, A., Jung, K. and López García, Á. Metric-privacy-inspired noise calibration in federated learning: Improving convergence and preventing client inference attacks. Knowledge-Based Systems 343, 115993 (2026). https://doi.org/10.1016/j.knosys.2026.115993.
- Kibsgaard, E., Jwa, A. S., Markiewicz, C. J., Rodriguez Gonzalez, D., Sáinz-Pardo, J., Poldrack, R. A., Pernet, C. R. (2026). Assessing metadata privacy in neuroimaging. Imaging Neuroscience 2026; 4 IMAG.a.1144. https://doi.org/10.1162/IMAG.a.1144.
- Sáinz-Pardo Díaz, J., & López García, Á. (2025). Enhancing the
Convergence of Federated Learning Aggregation Strategies with Limited Data. In 2025 3rd International Conference on Federated Learning Technologies and Applications (FLTA) (pp. 9-16). IEEE. https://doi.org/10.1109/FLTA67013.2025.11336682. - Dora, J. R., Hluchý, L., & Staňo, M. (2025). In-Memory Shellcode Runner Detection in Internet of Things (IoT) Networks: A Lightweight Behavioral and Semantic Analysis Framework. Sensors, 25(17), 5425. https://doi.org/10.3390/s25175425.
- Palacio Hoz, A., Heredia Canales, A.,Cimadevilla Álvarez, E., Obregón Ruiz, M., & López García, Á. (2025). DataLab as a service: distributed computing framework for multi-interactive analysis environments. IEEE Access, 13, 22566-22577. https://doi.org/10.1109/ACCESS.2025.3536637.
- Ascione, C., Valdano, E. How floods may affect the spatial spread of
respiratory pathogens: the case of Emilia-Romagna, Italy in May 2023.
EPJ Data Sci. 14, 43 (2025). https://doi.org/10.1140/epjds/s13688-025-00560-y. - Kelley, K., Gozzi, N., Mazzoli, M. et al. Exploring influenza vaccination determinants through digital participatory surveillance. BMC Public Health 25, 1345 (2025). https://doi.org/10.1186/s12889-025-22496-8.
- Fiandrino, Stefania, et al. “Collaborative forecasting of influenza-like illness in Italy: The Influcast experience.” Epidemics 50 (2025): 100819. https://doi.org/10.1016/j.epidem.2025.100819.
- Casaburi, Pasquale, et al. “Resilience of mobility network to dynamic population response across COVID-19 interventions: Evidences from Chile.” PLOS Computational Biology 21.2 (2025): e1012802. https://doi.org/10.1371/journal.pcbi.1012802.
- Sáinz-Pardo Díaz, J., López García, Á. An Open Source Python Library for Anonymizing Sensitive Data. Sci Data 11, 1289 (2024). https://doi.org/10.1038/s41597-024-04019-z.
- Liu, E., Mazzoli, M., Yan, XY. et al. A generalized vector-field framework for mobility. Commun Phys 7, 190 (2024). https://doi.org/10.1038/s42005-024-01672-z.
anjana
Overview:
Anjana is a Python library for data anonymization. It allows to apply the same anonymization techniques that pyCANON, allowing users to anonymize sensitive tabular data using such methods.
Designed to work in tandem with pyCANON, anjana it is especially suited for preparing datasets for public sharing while minimizing information loss.
Anjana has its own standalone service, and it is also integrated within the SIESTA Dashboard.
Resources:
SIESTA Dashboard
Overview:
The SIESTA Dashboard provides a user-friendly interface to explore, monitor and access the services and tools of the EOSC-SIESTA project. It is designed to facilitate researchers and stakeholders in interacting with the platform.
Resources:
Epidemiradar
Epidemiradar is a citizen science platform. Citizens voluntarily report their weekly health status via website, generating anonymised data displayed through interactive maps.
Resources:
pyCANON
Overview:
pyCANON is a Python library and command-line interface designed to analyze the anonymity of datasets. It implements a variety of privacy-preserving models, including:
k-anonymity
(α,k)-anonymity
ℓ-diversity (and entropy/recursive variants)
t-closeness
basic/enhanced β-likeness
δ-disclosure privacy
It outputs detailed reports for each model and helps data scientists, researchers, and policymakers evaluate the privacy level of tabular datasets.
Resources:
Paper in Scientific Data (Nature)
BidScramble
Overview:
BidScramble is an open-source tool developed at the Donders Institute. It provides methods to anonymize neuroimaging data in BIDS (Brain Imaging Data Structure) format, ensuring privacy while maintaining research usability.
Resources:
This Python build tool enables a given user to calculate a variety of different data privacy metrics on tabular data from a user interface.
Resources:
DatLeak
The purpose of this repository is to analyze information leakage in two different data types of NeuroImaging and Tabular datasets. Each of which consist of Original and Scrambled/Synthetic versions.
Resources:
TrasgoDP implements a set of mechanisms for Local Differential Privacy (LDP). It is particularly well-suited for generating synthetic versions of a dataset using mechanisms that also ensure differential privacy.
Resources: