Curriculum Vitae
Heidelberg, Germany
Research Interests
My research interests lie in exploring the nature of life and intelligence - both for "real" and artificial life. I'm especially interested in emerging properties of complex living systems, such as evolution, repair and self-organization.
In context of medical research, I'm interested in working with large-scale datasets and integration of various data modalities, with a focus on real-world applications beyond curated datasets.
In my free time, I like to explore deep learning as a tool for creativity.
Education
Ph.D. in Computer Science
Heidelberg University • October 2021 - present
Master's-level Program in CS and ML
Yandex School of Data Analysis • September 2022 - May 2024
BS/MS Applied and Fundamental Physical Chemistry
Moscow State University • September 2015 - August 2021
Research & Work Experience
Visiting Scientist, Birney Group
European Bioinformatics Institute (EBI) • May 2024 - present
- Designed a custom transformer model that estimates future disease risk based on previous medical history, surpassing the performance of traditional clinical score-based risk models
- Led the team during the project, coordinated manuscript preparation and created all illustrations
Doctoral Researcher, Computational Histopathology
Division of AI in Oncology (German Cancer Research Center) • October 2021 - May 2025
- Collaborated with institutions across seven countries to develop a transformer-based SOTA model for brain tumour classification, directly targeting clinical application
- Improved performance of tumour classification models by utilising SSL methods (DINO, iBOT), fine-tuned on in-house data
- Identified the main causes of performance drop when validating on external cohorts
Research Intern, Cancer Evolution Group
European Bioinformatics Institute (EBI) • June 2019 - September 2019
- Mathematically formulated, implemented, and maintained a popular cell type decomposition package (cell2location, 300+ GitHub stars)
- Developed a Bayesian model for spatial cancer genomics data using Gaussian Processes
- Implemented a cell nuclei segmentation pipeline suitable for ultra-large images (up to 100k x 100k px)
Selected Publications
Learning the natural history of human disease with generative transformers
Shmatko, A., Jung, A.W., Gaurav, K., Brunak, S., Mortensen, L.H., Birney, E., Fitzgerald, T., Gerstung, M.
Nature • 2025 • DOI: 10.1038/s41586-025-09529-3 Paper • Code
Artificial intelligence in histopathology: enhancing cancer research and clinical oncology
Shmatko, A., Ghaffari Laleh, N., Gerstung, M. et al.
Nature Cancer • 3, 1026–1038 • 2022 Paper
Comprehensive mapping of tissue cell architecture via integrated single cell and spatial transcriptomics
Kleshchevnikov, V., Shmatko, A., Dann, E. et al.
Talks & Conferences
Computational histopathology enables high-granularity diagnostics in CNS tumours
The 19th Meeting of the European Association of Neuro-Oncology • Glasgow, UK • 2024 • Oral presentation
Delphi: learning the natural history of human disease with generative transformers
Bristol University Health Seminar Series • Bristol, UK • 2024 • Oral presentation Video