Artificial Intelligence in Oncology - Supporting scientific research
Radboud UMC
In short, Iris Nagtegaal's research entails the following:
Cancer of unknown primary sites (CUP) refer to all metastatic cancers where no primary location has been determined. This patient group accounts for 2.5% of all newly diagnosed cancer patients. The outcomes in this patient group are poor, because treatment is usually based on the primary location of the cancer. Even with the current improved diagnostic methods, ever year approximately 1,300 patients are diagnosed with CUP in the Netherlands. Their survival is less than 6 months. With artificial intelligence, we will explore big data in order to determine metastatic patterns of patients with a known primary location and develop a probability model based on metastatic location. We will subsequently incorporate routinely used biomarkers to optimize decision making, which we will test in a proof-of-concept study. As a result of our study we will deliver a prediction model that can be used in a prospective clinical trial to improve outcomes of patients with CUP.'
After the collection of data and material in the first two years, we have worked very hard this year on the analytical phase, exploring different types of machine learning approaches and subsequently testing and validating these. Since this took more time than expected, we have continued with the study of rare metastases and have initiated several ways of biomarker collections and explorations.
During the first two years of the project, we have focussed on data collection and method development. We have collected the relevant datasets that are necessary to progress and are currently comparing the adequacy of different machine learning techniques for the tasks we have set. At the European Congress of Pathology we presented our first results on breast cancer metastases.
In the first year of the project, we have collected and annotated large patient series that form the basis of our project. We started with the design of the AI approach to our databases and obtained the first results on metastatic patterns in breast and lung cancer patients. In addition, we have performed a global Delphi consensus study into the diagnostic workup of CUP and assisted in the design of a systematic review into CUP risk factors.
On World CUP awareness day (September 24th, 2021) we presented this project online int he Update on CUP inititatives & research in the Netherlands.