First publication from APPROACH
An article describing part of the computational science behind the APPROACH project has been published in the Scientific Reports journal. It is a result of several years of research and hard work, and we are very proud of it!
Title
Multi-classifier prediction of knee osteoarthritis progression from incomplete imbalanced longitudinal data
Authors
Paweł Widera, Paco M. J. Welsing, Christoph Ladel, John Loughlin, Floris P.J.G. Lafeber, Florence Petit Dop, Jonathan Larkin, Harrie Weinans, Ali Mobasheri & Jaume Bacardit
Available here.
Summary
In APPROACH, machine learning models have been used to decide which patients to include in the study. Our research hypothesis was that these models could more precisely select patients, who are likely to experience the OA progression (compared to conventional criteria).
The recently published article, describes computational experiments performed on the OA data in order to find the best way to learn how to predict different kinds of progression. Step by step, the article evaluates multiple algorithms and learning processes, and investigates the impact of different input information on the best models behaviour.
Finally, it looks at how good these models are in a simulated patient selection scenario. The models were found to choose clinically more desirable group of patients, and with modifications, have been later made a part of the APPROACH recruitment process.
