Machine learning models for inclusion of progressors in OA clinical studies

by PaweĊ‚ Widera

16:00 (40 min) in USB 2.022

To study complex diseases, we need to observe how they progress over time. For that purpose, we set up clinical studies in which over several months, or even years, we monitor changes in the patients health. The success of such a study depends heavily on the inclusion procedure. If we don't include enough patients with observable disease progression, the collected data will not tell us much about the disease development.

As part of the APPROACH project, a multi-centre, 2-year follow-up, exploratory study of knee osteoarthritis (OA) funded by IMI/EU, we worked on improving the inclusion process using machine learning. In this talk, I will describe how we built the machine learning models using data from 5 European hospitals, how we designed a two-stage recruitment process around the model predictions, and finally, how we overcame a number of practical, data-related challenges.