Prediction of pain in knee osteoarthritis patients using machine learning: Data from Osteoarthritis Initiative

摘要

Knee Osteoarthritis(KOA) is a serious disease that causes a variety of symptoms, such as severe pain and it is mostly observed in the elder people. The main goal of this study is to build a prognostic tool that will predict the progression of pain in KOA patients using data collected at baseline. In order to do that we leverage a feature importance voting system for identifying the most important risk factors and various machine learning algorithms to classify, whether a patient’s pain with KOA, will stabilize, increase or decrease. These models have been implemented on different combinations of feature subsets, and results up to 84.3% have been achieved with only a small amount of features. The proposed methodology demonstrated unique potential in identifying pain progression at an early stage therefore improving future KOA prevention efforts.