
IICU National Data Stream
Personalized, data-driven prediction and assessment of infection-related outcomes in Swiss intensive care units (ICUs)

Abstract
The IICU (Personalized, data-driven prediction and assessment of Infection related outcomes in Swiss Intensive Care Units) National Data Stream (NDS) aims to improve the care of critically ill patients with infections across Switzerland. It brings together detailed clinical and pathogen data from multiple sources to better understand how infections develop, why treatments are chosen, and how patients respond.
By connecting medical experts and data scientists, the project seeks to better understand individual patient conditions, predict complications earlier, and support more personalized treatment decisions, including the appropriate use of antibiotics. The IICU NDS aims to make its data accessible and to publish tools and models that help improve patient outcomes in intensive care and advance infection research.
This project has been funded by SPHN and PHRT from 2022 to 2026.
People
Collaborators


Xinrui obtained her Ph.D. in Computer Science from ETH Zürich in 2022 under the supervision of Prof. Gunnar Rätsch. During her doctoral study, she was devoted to applying machine learning algorithms to solving healthcare problems. Xinrui also holds a M.Sc. degree in Electrical Engineering from EPFL in Switzerland and a B.Eng. degree in Electronic Information Science and Technology from Tsinghua University in China. Before joining the SDSC, Xinrui worked for the NEXUS Personalized Health Technologies platform at ETH Zürich.


Irene joined the SDSC in August 2024 as a Senior Biomedical Data Scientist. Trained as an evolutionary biologist, she worked in academic research in Switzerland and abroad before transitioning to bioinformatics, where she specialized in supporting researchers with the analysis of next generation sequencing data. At SDSC, her focus is on data engineering, management, and analysis of biomedical data.


Nora earned her Ph.D. in Computer Science/Bioinformatics from the University of Tübingen, where she focused on the in silico design of peptide-based vaccines using combinatorial optimization and machine learning. During her postdoctoral fellowship at Memorial Sloan Kettering Cancer Center in New York, and later as a staff scientist at the New York Genome Center, she worked on metagenomics, infectious diseases, and cancer. In 2015, Nora joined NEXUS Personalized Health Technologies at ETH Zurich where her focus shifted towards the management of clinical and biomedical research data. In May 2024, she joined the Swiss Data Science Center as Head of Biomedical Data Science.
PI | Partners:
Project partners:
- Centre Hospitalier Universitaire Vaudois (CHUV)
- Hôpitaux Universitaires de Genève (HUG)
- Insel Gruppe AG, University Hospital Bern (Insel)
- University Hospital Basel (USB)
- University Hospital Zurich (USZ)
- Swiss Pathogen Surveillance Platform (SPSP)
- University of Geneva
- University of Bern
- University of Basel
- University of Zurich (UZH)
- University of Lausanne
- ETH Zürich (ETHZ)
- SIB Swiss Institute of Bioinformatics
Main PIs:
Prof. Adrian Egli (UZH),
Prof. Catherine Jutzeler (ETHZ),
Prof. Karsten Borgwardt (ETHZ, until 1 February 2023)
Project consortium:
Prof. Manuel Battegay (USB),
Prof. Thierry Calandra (CHUV),
Prof. Jean-Daniel Chiche (CHUV),
Prof. Hansjakob Furrer (Insel),
Prof. Gilbert Greub (CHUV),
Prof. Stephan Jakob (Insel),
Prof. Catherine Jutzeler (ETHZ),
Dr. André Kahles (ETHZ),
Prof. Laurent Kaiser (HUG),
Dr. Aitana Lebrand (SIB),
Dr. Stephen Leib (Insel),
Prof. Sylvain Meylan (CHUV),
Prof. Jerome Pugin (HUG),
Prof. Yok-Ai Que (Insel),
Prof. Gunnar Rätsch (ETHZ),
Prof. Thierry Roger (CHUV),
Prof. Jacques Schrenzel (HUG),
Prof. Reto Schüpbach (USZ),
Prof. Martin Siegemund (USB),
Prof. Reinhard Zbinden (USZ),
Prof. Annelies Zinkernagel (USZ)
Patient partner:
Dominique Loosli
Project manager:
Dr. Magdalena Lukamowicz-Rajska (USZ, until 31 June 2026)
Data manager:
Dr. Xinrui Lyu (SDSC),
Dr. Irene Keller (SDSC)
description
Objectives
The IICU National Data Stream is a Swiss multi-institutional initiative to better understand the complex dynamics between intensive care patients, the microorganisms causing infections, and the timing of treatment.
Building on the Personalized Swiss Sepsis Study (PSSS), it combines clinical, contextual, and pathogen data to improve personalized assessment, characterization, and outcome prediction in patients with infections, taking into account the attending physicians’ clinical reasoning and their resulting patient assessments.
This knowledge, which goes beyond routine clinical records, substantially improves the quality of the cohort data and forms a globally unique research data repository for the benefit of critically ill patients with infections.
Approach
The project aims to improve personalized assessment, treatment, and outcome prediction for patients with severe infections by combining clinical expertise with data science and AI.
To achieve this, data from over 90,000 intensive care unit (ICU) patients across Switzerland’s five university hospitals have been collected and harmonized. The repository integrates clinical records, continuous patient monitoring, laboratory results, medication data, infection diagnostics, and pathogen omics processed via the Swiss Pathogen Surveillance Platform (SPSP).
This comprehensive, privacy-preserving data resource enables researchers to identify digital biomarkers, predict infection and related outcomes, better understand infection types and treatment responses, and develop more personalized approaches to intensive care.
SDSC Contributions
The data scientists and engineers at the Swiss Data Science Center (SDSC) have established the technical and data foundations for the IICU National Data Stream, enabling the secure, consistent, and effective use of clinical data across participating hospitals.
Working closely with clinicians, hospital data teams, and other partners, the SDSC supports the entire data lifecycle, from secure integration and quality assurance to standardization, anonymization, and controlled sharing. This includes harmonizing complex datasets from multiple institutions and aligning them with national standards.
Specific tasks carried out by the SDSC team include:
- Data management: decrypting incoming data, onboarding new users to the system, and supporting researchers throughout their work with the data.
- Cross-stakeholder alignment: coordinating with hospital clinical data warehouses, clinicians, and the SPHN Data Coordination Center (DCC).
- Project-specific concept schema design: developing schemas tailored to the project’s needs while ensuring alignment with SPHN concepts and standards.
- Data quality checks: reviewing datasets for inconsistencies, artifacts, and gaps before further analysis.
- Data preprocessing and harmonization: preparing and aligning datasets from different hospitals so that they are consistent and ready for machine learning.
- Data anonymization: applying anonymization methods that protect patient privacy while preserving the analytical value of the data.
- Data sharing: making the data available to third-party users under restricted access to promote reproducible research in the infectious disease area.
These contributions provide the basis for collaborative, reproducible research while safeguarding patient privacy and supporting the development of data-driven approaches to improve infection-related care.
Projected Impact
The IICU National Data Stream aims to improve intensive care for patients with severe infections while delivering broader benefits to society:
- Help reduce mortality: Enable earlier detection of infection and more personalized treatment through AI-driven predictions.
- Preserve antibiotic effectiveness: Support more targeted antibiotic use, reducing the risk of drug-resistant bacteria.
- Advance medical research: Provide third-party researchers with secure access to high-quality ICU data.
- Strengthen Switzerland’s healthcare system: Establish reusable data infrastructure and tools that support future research, innovation, and more effective patient care.
Presentation
Gallery
Annexe
Additional resources
Bibliography
Publications
Related Pages
- SPHN project page: https://sphn.ch/network/projects/project-page_nds_iicu/
- PHRT project page: https://sfa-phrt.ch/projects/clinical/personalized-data-driven-prediction-and-assessment-of-infection-related-outcomes-in-swiss-icus-iicu/
- IICU kick-off presentation: https://sphn.ch/wp-content/uploads/2022/10/NDS-2021-902_KickOffPresentation.pdf
- The IICU NDS RDF Schema: https://www.biomedit.ch/rdf/sphn-schema/iicu
- SPHN Metadata Catalog: https://fdp.dcc.sib.swiss/catalog/64136f6d-1e2b-557d-8c9d-5704e7c4c2e9
More projects
More projects
News
Latest news
Elevating International Geneva's Impact with AI
Elevating International Geneva's Impact with AI
12 Zurich SMEs join next AI Innovation Program
12 Zurich SMEs join next AI Innovation Program
La science des données au service de la neutralité carbone du canton : ENERBAT ouvre la voie
La science des données au service de la neutralité carbone du canton : ENERBAT ouvre la voie
Contact us
Would you like to explore how we can work together?
Our team of data scientists and professionals looks forward to engaging in impactful, results-driven collaborations with you.



