Seminal
The magazine and atlas of EU-funded researchA new publication from the web and AI team at Scott Charnley

Atlas / Switzerland

Marie Skłodowska-Curie Action · Running

OUTBRAID OUTBreak AI-driven Detection: Enhancing MALDI-TOF outbreak detection with multimodal AI integrating epidemiological and genomic data

Coordinator
Universitat Zurich, Switzerland
Runs
1 July 2026 to 30 June 2028
Status
Running
EU funding
€308k
Total cost
—
Partners
2 organisations in 2 countries
Type
Marie Skłodowska-Curie Action
Grant agreement
101278403

What the project is doing

In the project's own words, from its summary on CORDIS.

Hospital-acquired infections (HAIs) from resistant and hypervirulent bacteria—such as Methicillin-resistant Staphylococcus aureus, Vancomycin-resistant Enterococcus faecium, Carbapenemase-producing Pseudomonas aeruginosa, Carbapenemase-producing Acinetobacter baumannii, and Clostridioides difficile—pose a growing threat to patient safety worldwide. Outbreaks often go undetected until severe harm occurs. While whole genome sequencing (WGS) is the gold standard for outbreak tracing, it is costly, slow and not globally accessible.

MALDI-TOF mass spectrometry, in contrast, is rapid and widely used, but its outbreak detection potential remains unexploited. The OUTBReak AI-driven Detection (OUTBRAID) project will transform hospital outbreak surveillance by leveraging an unprecedented international dataset—over 1.2 million MALDI-TOF spectra already collected at the IMM Zurich, plus new contributions from 9 countries across Europe, Asia, and Australia, including Low and Median Income Countries (LMICs), each providing at least 100 samples per pathogen for rigorous, external validation.

OUTBRAID will develop two breakthrough approaches: (1) diffusion models to generate synthetic MALDI-TOF spectra, ensuring robust, cross-center performance; and (2) multimodal machine learning that combines MALDI-TOF, epidemiological metadata, and WGS to enable automated, high-resolution cluster detection. These models will be benchmarked not only at species level but also at sequence type (ST), cgMLST, and single-nucleotide polymorphism (SNP) resolution—pushing MALDI-TOF as close as possible to the granularity of WGS.All data, models, and code will be open access.

OUTBRAID’s participation in workshops and training will reach and train 150+ scientists and clinicians (e.g., ESCMID), ensuring broad impact. By braiding together international data streams and AI, OUTBRAID aims to deliver scalable tools for early outbreak detection, supporting hospitals and public health globally.

Topics

Who is involved

OrganisationCityCountryRole
Universitat ZurichZurichSwitzerlandCoordinator
Max-Planck-Gesellschaft Zur Forderung Der Wissenschaften eVMunchenGermanyAssociatedpartner

Source: CORDIS, European Commission. Data of 5 Nov 2026, reused under the Commission's reuse policy. View on CORDIS · Grant DOI