CORAL - Community in Oncology for RApid Learning


(new name for euroCAT/sinoCAT/ozCAT/ukCAT/VATE/meerCAT etc.)

  • CORAL (Community in Oncology for Rapid Learning) is a community of 20-30 cancer centers worldwide (Netherlands, USA, UK, India, China, South-Africa, Australia, Italy, Germany, Belgium, Canada, Denmark)
  • Initiated form Maastricht
  • Radiation oncology focus and mostly in lung, rectum and head&neck cancer
  • Structured and imaging data as input, unstructured data is a challenge esp. globally
  • Distributed approach (data does not cross the firewall) where A) data is first made semantically interoperable locally (nowadays called FAIR) and B) centers allow applications to enter their firewall and use their data to answer questions
  • Applications are focused on machine learning / modelling (a logistic regression model to predict overall survival in lung cancer)
  • MAASTRO and others in CORAL are developing “A)” an open source suite to make your data FAIR
  • Varian Medical Systems is developing “B)” the distributed application environment
  • Ontologies used are NCI Thesaurus, FMA, Unit Ontology. ROO (Radiation Oncology Ontology) is maintained for concepts which cannot be found somewhere else
  • Challenges are industrial grade software needed to scale up


  • Gearing up for a demo at ASCO 2016.
  • Centralized approach with 30-40 USA centers currently focused on breast cancer and medical oncology. Discussing a possible collaboration with ASTRO.
  • Challenges are some centers having policies not to send data outside their firewall. Europe restricts data flows to USA.
  • Free, unstructured text is a challenge, often important cancer outcomes are hard to extract
  • CancerLinQ in a box is a medium term aim, this would decentralize it, enable centers to share data bilaterally
  • SAP and ASCO are partner in a CancerLinQ foundation
  • Type of questions are currently mostly statistical, aggregate (how many patients with these characteristics are you treating)
  • NCI Thesaurus is the base ontology

Possible collaborative projects

  • Send someone from Maastricht over to set up CancerLinQ as a “node” in CORAL, show we can combine forces and learn e.g. a prediction model for breast
  • Establish a prototype CancerLinQ-in-a-Box at Maastricht by combining some of the technologies/approaches of CancerLinQ and CORAL. Aim is to show you can join CancerLinQ in a decentralized manner, in Europe etc.
  • Work with Foundation/SAP top make the extraction CORAL part “A)” industrial grade and leverage that inside CancerLinQ



  • Wilkinson MD, Dumontier M, Aalbersberg IjJ, et al: The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data 3:160018, 2016

CAT/Rapid Learning:

  • Lambin P, Zindler J, Vanneste B, et al: Modern clinical research: How rapid learning health care and cohort multiple randomised clinical trials complement traditional evidence based medicine. Acta Oncologica 1–12, 2015
  • Lambin P, Zindler J, Vanneste BGL, et al: Decision support systems for personalized and participative radiation oncology [Internet]. Advanced Drug Delivery Reviews , 2016[cited 2016 Feb 19]

Tech used: see att.


Attendee list CORAL meeting ESTRO 35.pdf


CORAL photo