About the Project

The research project "FAIrPaCT" is funded by the German Federal Ministry of Education and Research as part of the Decade Against Cancer, and aims to develop a software system that can predict the probability of success of individual therapies against pancreatic cancer. This should identify factors which molecular mechanisms control a therapy and increase the chances of effective treatment. Insights that can lead to improved drugs and personalized treatment strategies. It would be a big step in the overcoming pancreatic cancer, which is considered one of the deadliest types of cancer. Only seven to eight percent of those affected survive the first five years after diagnosis. The tumors are often only detected late and form metastases.

"FAIrPaCT" stands for "Framework for federated artificial intelligence for the optimization of pancreatic cancer treatment". The system that the researchers want to develop is based on a so-called federated ensemble model. This is an algorithm that enables machine learning with data sources from different locations. In addition, the data is not centrally stored, but is stored on various sites. The researchers from FAIrPaCT can therefore analyze data across institutes without centrally storing it. This ensures higher data protection, which is particularly important for patient data. "FAIrPaCT" uses three of the largest and most meaningful data sets of patients with pancreatic cancer in Germany. These are clinical patient data and molecular data from removed cancer cells. These data were obtained from the clinics of the consortium: at the University Medical Center Göttingen, at the University Hospital Gießen and Marburg, and at the Klinikum Rechts der Isar of the Technical University of Munich. The amount and diversity of the data is unique throughout Germany.


FAIrPaCT“ steht für „Framework für föderierte künstliche Intelligenz zur Optimierung der Behandlung von Bauchspeicheldrüsenkrebs“. Das System, das die Forschenden entwickeln wollen, basiert auf einer sogenannten föderierten oder verteilten künstlichen Intelligenz. Das ist ein Algorithmus, der maschinelles Lernen mit Datenquellen von unterschiedlichen Standorten ermöglicht. Zudem werden die Daten nicht zentral ablegt, sondern auf verschiedenen Geräten gespeichert und sind flexibel abrufbar. Die Forschenden von FAIrPaCT können Daten also institutsübergreifend analysieren, ohne sie zentral abzulegen. Damit ist ein höherer Datenschutz gewährleistet, was insbesondere bei Patientendaten eine wichtige Rolle spielt. „FAIrPaCT“ nutzt drei der größten, aussagekräftigsten Datensätze von Patientinnen und Patienten mit Bauchspeicheldrüsenkrebs in Deutschland. Dabei handelt es sich um klinische Patientendaten und molekulare Daten von entnommenen Krebszellen. Gewonnen wurden diese Daten in den Kliniken des Verbundes:  in der Universitätsmedizin Göttingen, im Uniklinikum Gießen und Marburg sowie im Klinikum Rechts der Isar der Technischen Universität München. Die Menge und Verschiedenartigkeit der Daten ist deutschlandweit einzigartig.


Project Goals

Artificial intelligence for better therapies. Together, the research team will analyze data stored in different sites. Based on this analysis, medical informaticians will develop a software system that can estimate the probability of success for specific treatment approaches. Since the data comes from three different locations, the calculations go beyond the boundaries of one location. Therefore, the system will be usable regardless of location in the future. In addition, it should identify important parameters that influence the response to a particular treatment. By using the data collection from all three locations, the researchers aim to find general factors for the success of therapy against pancreatic cancer. This work is an important step towards precision medicine supported by artificial intelligence.

In the long term, patients with pancreatic cancer, as well as oncologists, are expected to benefit from the data analysis. The analyzed data can help in making treatment decisions and potentially increase the survival chances of patients with pancreatic cancer. In addition, the participants of the research project "FAIrPaCT" aim to better understand the development of tumors in the pancreas and the molecular mechanisms that lead to the success or failure of treatment. This will help create the conditions for the development of new personalized therapies and demonstrate that the use of software systems based on federated AI can significantly support both science and clinical practice. The federated learning methods developed in "FAIrPaCT" can also be used for other cancer types in the future.

"FAIrPaCT" is one of eight research consortia that conduct research within the framework of BMBF funding for interdisciplinary projects to develop and test new approaches to data analysis and sharing in cancer research during the Decade Against Cancer. The aim of this funding measure is to provide more data analysis researchers with low-threshold access to existing cancer research and oncological routine care data. The application of newly developed data analysis approaches is expected to help filter and utilize research-relevant information. At the same time, the data sharing culture for research purposes should be promoted. Thus, this research cases can also serve as best-practice solutions for joint data utilization in cancer research.




  • Technische Universität München, Klinikum rechts der Isar, Clinic and Polyclinic for Internal Medicine II, Dr. Maximilian Reichert



Deputy Head / W1 professorship Clinical Decision Support

Prof. Dr. Anne-Christin Hauschild

Clinical decision support

Dr. Youngjun Park

Follow us