Radiological pain predictors in knee osteoarthritis, a four feature selection comparison: Data from the OAI

Abstract

In medical science, the image based biomarkers are a recent tool for disease diagnostic and prognostic, the different medical imaging techniques brings a big amount of useful data for analysis and interpretation. The osteoarthritis (OA) is a very common and disabling disease in the industrialized world, pain is the most important and disabling symptom in knee OA, having a preventive treatment is one of the most important tasks in the OA studies. In this work a bioinformatic tool is used to obtain pain prediction models, using genetic algorithms with different feature selection functions multivariate prediction models were obtain and compared based on the medical requirements to investigate radiological features that precede the onset of knee pain, and to identify a radiological-based multivariate prognostic model of knee pain. © 2014 Springer International Publishing.