idw – Informationsdienst Wissenschaft
Nachrichten, Termine, Experten
Nachrichten, Termine, Experten
idw - Informationsdienst
Wissenschaft
Researchers from University Hospital Cologne have developed an autonomous agent-based AI system called ‘SPARK’ that acts as a “digital brain” / publication in ‘Nature Medicine’
The digital transformation of pathology is opening up new possibilities for cancer diagnosis. Today’s artificial intelligence (AI) techniques now go far beyond mere automation: they make it possible to extract previously hidden biological information from routinely collected histological tissue sections, thereby enabling a deeper understanding of tumour diseases.
A team from the Department of Pathology at University Hospital Cologne has developed a fundamentally new approach called SPARK (System of Pathology Agents for Research and Knowledge). The study “An agentic framework for autonomous scientific discovery in cancer pathology” was published in the journal Nature Medicine.
Conventional AI approaches primarily focus on segmenting tissue or analysing individual cells within the tumour microenvironment. However, they often reach their limits – for example, due to limited interpretability or a lack of transferability to new research questions. The agent-based AI system acts as a “digital brain”. It connects several specialized algorithms to a coordinated system that can generate autonomously biological hypotheses, improve them, and translate them into analytical tools without the need to retrain the models. The use of language as a universal interface enables flexible and intuitive interaction with complex image data. This makes it possible, for example, to carry out simple language-based analyses, such as determining whether a tumour will respond to immunotherapy.
In extensive analyses of over 5,400 patients from 18 independent cohorts and five different tumour types, the team led by Dr Yuri Tolkach, Senior Physician at the Institute of Pathology at University Hospital Cologne, demonstrated that SPARK identifies clinically relevant and biologically grounded tissue markers. These are closely linked to the course of the disease, established pathological parameters, and the response to treatment.
Furthermore, the system enables conclusions to be drawn from static tissue sections regarding the temporal development of tumours and provides a better understanding of the mechanisms underlying tumour progression.
“SPARK helps to refine diagnoses, stratify patients more reliably, and make more precise treatment decisions. Particularly in the field of personalized oncology, there is an opportunity to tailor treatments more closely to the individual biological characteristics of a tumour, thereby improving treatment outcomes,” says Tolkach.
According to the researchers, the accessibility of SPARK is another advantage: via a specialized, interactive modular user interface, clinicians and researchers can develop analytical approaches even without any programming knowledge.
Despite these promising results, prospective validation in everyday clinical practice is necessary to confirm the full benefits of the technology. All the methods, parameters, and results developed have been made openly available in order to actively encourage further development by the academic community. “With SPARK, we aim to transform pathology from a primarily descriptive discipline into a data-driven, predictive science – and thereby make a significant contribution to precision medicine in oncology,” says Professor Dr Reinhard Büttner, Director of the Institute of General Pathology and Pathological Anatomy.
SPARK was funded, amongst others, by the former German Federal Ministry of Education and Research (BMFTR) and as part of the DigiPathConnect project under the European Union’s Interreg Euregio Meuse-Rhine programme. In addition, data from the National Network for Genomic Medicine in Lung Cancer (nNGM, funded by German Cancer Aid) were used, as was the computing power of the RAMSES supercomputer at the IT Center University of Cologne (ITCC).
PD Dr Yuri Tolkach
Institute of Pathology at University Hospital Cologne
+49 221 478 6365
iurii.tolkach@uk-koeln.de
https://www.nature.com/articles/s41591-026-04357-y
Merkmale dieser Pressemitteilung:
Journalisten, Studierende, Wissenschaftler
Informationstechnik, Medizin
überregional
Forschungsergebnisse, Wissenschaftliche Publikationen
Englisch
Sie können Suchbegriffe mit und, oder und / oder nicht verknüpfen, z. B. Philo nicht logie.
Verknüpfungen können Sie mit Klammern voneinander trennen, z. B. (Philo nicht logie) oder (Psycho und logie).
Zusammenhängende Worte werden als Wortgruppe gesucht, wenn Sie sie in Anführungsstriche setzen, z. B. „Bundesrepublik Deutschland“.
Die Erweiterte Suche können Sie auch nutzen, ohne Suchbegriffe einzugeben. Sie orientiert sich dann an den Kriterien, die Sie ausgewählt haben (z. B. nach dem Land oder dem Sachgebiet).
Haben Sie in einer Kategorie kein Kriterium ausgewählt, wird die gesamte Kategorie durchsucht (z.B. alle Sachgebiete oder alle Länder).
