SAGE

There is a growing evidence gap in personalized cancer treatment decisions, likewise in radiotherapy. Issues include the rapid development and implementation of novel technologies and treatments, the difficulties of comparing these in randomized clinical trials (RCT) in radiotherapy (RT) and the unrepresentative selection criteria used in RCTs. However, there is a huge amount of routine clinical data collected every day in every cancer center that is not used for evidence generation, yet is complementary to RCT data and could support optimized clinical decisions.

Combined routine clinical data from different cancer practices provide a larger, more relevant, dynamic and potentially stronger evidence base on which to base decisions, a research area which is now commonly called Rapid Learning. In the previous Varian-supported VATE project the possibility for wide access to large amounts of data shared by centers by using automated software agents in a distributed learning perspective was implemented. The SAGE project, aims to develop a framework for addressing medically relevant issues in data mining. This framework is applied on available data to support knowledge promotion and for detecting and correcting data quality issues to document the quality and acceptance of evidence from Rapid Learning. This initiative serves as an extension to the Rapid Learning infrastructure project VATE. The goal is to enhance confidence in outcome prediction model performance, countering the "garbage-in/garbage-out" argument.

This project was awarded funding by Varian in 2015 and was concluded in 2019.