Arrow Left Health research

Use case: chinaCAT: Rapid learning infrastructure for outcome prediction models in rectal cancer

Developing a rapid learning infrastructure for rectal cancer outcome prediction models.

Medical relevance

Personalized decision support for rectal cancer patients


chinaCAT established an infrastructure to share patient data from multiple international institutions. This infrastructure will help to overcome the current administrative, political, ethical and technical barriers. Radiation treatment of rectum cancer patient’s generate large amounts of data. In this project, we used machine learning based on shared data to develop outcome prediction models. These models will provide personalized guidance and decision support for treatment of rectal cancer.

Main results

New outcome prediction models developed by machine learning and validated using European and Chinese oncology treatment data.

Lessons learned

Distributed learning works and leads to meaningful clinical knowledge even when the data-sources are extremely different in terms of patient-population, language, culture and treatments.

Follow up

After the chinaCAT ended, Fudan University joined the 20k challenge. Maastricht University and Fudan University maintain active research collaborations including joint PhD students.

Project details

Project leader

Zhen Zhang

Funders Varian Medical Systems
Collaboration partners

Fudan University Shanghai Cancer Center, Shanghai, China / MAASTRO Clinic, Maastricht Netherlands / University of Pennsylvania, Philadelphia, PA, USA / Università Cattolica S Cuore, Rome, Italy



Validation of a rectal cancer outcome prediction model with a cohort of Chinese patients

Share this page…