Coordinated MELLODDY (largest EC funding at EUR 2.68M), which built privacy-preserving machine learning across pharma partners for predictive drug discovery models.
OWKIN FRANCE
French AI SME specializing in federated learning and machine learning for drug discovery, cancer diagnostics, and clinical data analysis.
Their core work
Owkin is a Paris-based AI company that applies machine learning — particularly federated learning and deep learning — to healthcare and drug discovery. They specialize in building predictive models from complex biomedical data (pathology images, multi-omics, clinical records) while preserving data privacy across institutions. Their core value lies in enabling hospitals and pharma companies to collaborate on AI models without sharing raw patient data. In H2020 projects, they contribute AI/ML expertise to cancer research, rare disease analysis, cardiac diagnostics, and digital pathology.
What they specialise in
Contributed to IMMUcan (cancer immunoprofiling), BIGPICTURE (digital pathology repository), and OPTIMA (AI for solid tumour treatment optimization).
Participated in BIGPICTURE (central repository for digital pathology) and MAESTRIA (cardiac imaging with AI-based diagnostics).
ImmunAID involved multiomics data integration for autoimmune disorders; IMMUcan combined RNAseq, CyTOF, and bioinformatics for cancer immunoprofiling.
Participated in ImmunAID, focusing on machine learning applied to microbiome and inflammasome data for autoinflammatory disorders.
How they've shifted over time
Owkin's early H2020 work (2018-2019) centered on applying ML to specific disease biology — microbiome analysis, immune-mediated inflammatory diseases, and cancer immunoprofiling with multi-omics data. By 2021, their focus shifted decisively toward AI infrastructure for medicine: digital pathology repositories, cardiac AI diagnostics, and real-world evidence analysis for treatment optimization. The trajectory shows a company moving from disease-specific ML applications toward building scalable AI platforms for broader clinical use.
Owkin is evolving from a specialized ML contributor into a platform-level AI partner for large-scale clinical data projects, making them increasingly relevant for any consortium needing privacy-preserving medical AI.
How they like to work
Owkin operates primarily as a specialist partner (5 of 6 projects), contributing AI/ML capabilities to large health consortia. They coordinated one major project (MELLODDY), demonstrating leadership capacity when federated learning is the core technology. With 156 unique partners across 24 countries, they function as a well-connected technical contributor rather than a repeat-partner organization — they bring AI expertise into diverse medical research teams.
Owkin has collaborated with 156 unique partners across 24 countries, indicating broad European reach and high demand for their AI capabilities across different medical research communities. Their network spans university hospitals, pharma companies, and research institutes involved in oncology, cardiology, and rare diseases.
What sets them apart
Owkin occupies a rare niche as an SME that combines deep machine learning research with practical healthcare deployment, particularly in privacy-preserving federated learning. Unlike academic AI labs, they bring production-grade ML infrastructure; unlike big tech, they are deeply embedded in clinical research consortia. Their coordination of MELLODDY — a multi-pharma federated learning project — demonstrates a unique ability to build trust across competing organizations for collaborative AI.
Highlights from their portfolio
- MELLODDYOwkin's only coordinator role and largest funding (EUR 2.68M) — a landmark federated learning project enabling multiple pharma companies to train shared drug discovery models without exposing proprietary data.
- BIGPICTUREA flagship EU project building Europe's central repository for digital pathology, positioning Owkin at the infrastructure layer of AI-driven cancer diagnostics.
- IMMUcanLarge-scale cancer immunoprofiling project running through 2026, combining deep tissue analysis with AI across multiple cancer types — shows Owkin's long-term commitment to oncology AI.