![[2602.10168] EVA: Towards a universal model of the immune system](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.10168] EVA: Towards a universal model of the immune system
Summary
The paper introduces EVA, a universal multimodal foundation model for immunology that integrates diverse biological data to enhance drug development and patient stratification.
Why It Matters
EVA represents a significant advancement in immunology research by providing a comprehensive model that harmonizes various data types across species. This could lead to better understanding and treatment of immune-mediated diseases, addressing a critical gap in current biological models that often overlook complex multicellular interactions.
Key Takeaways
- EVA is the first cross-species, multimodal model for immunology.
- It integrates transcriptomics and histology data for improved patient representations.
- The model demonstrates state-of-the-art results across 39 drug development tasks.
- Clear scaling laws show that larger models improve performance in pretraining and downstream tasks.
- An open version of EVA is released to accelerate research in immune-mediated diseases.
Quantitative Biology > Quantitative Methods arXiv:2602.10168 (q-bio) [Submitted on 10 Feb 2026 (v1), last revised 12 Feb 2026 (this version, v2)] Title:EVA: Towards a universal model of the immune system Authors:Scienta Team: Ethan Bandasack, Vincent Bouget, Apolline Bruley, Yannis Cattan, Charlotte Claye, Matthew Corney, Julien Duquesne, Karim El Kanbi, Aziz Fouché, Pierre Marschall, Francesco Strozzi View a PDF of the paper titled EVA: Towards a universal model of the immune system, by Scienta Team: Ethan Bandasack and 10 other authors View PDF HTML (experimental) Abstract:The effective application of foundation models to translational research in immune-mediated diseases requires multimodal patient-level representations that can capture complex phenotypes emerging from multicellular interactions. Yet most current biological foundation models focus only on single-cell resolution and are evaluated on technical metrics often disconnected from actual drug development tasks and challenges. Here, we introduce EVA, the first cross-species, multimodal foundation model of immunology and inflammation, a therapeutic area where shared pathogenic mechanisms create unique opportunities for transfer learning. EVA harmonizes transcriptomics data across species, platforms, and resolutions, and integrates histology data to produce rich, unified patient representations. We establish clear scaling laws, demonstrating that increasing model size and compute translates to improvements in both...
