Abstract

sKnee osteoarthritis is a widespread problem among adults globally, with no current medications to cure it. Early detection is crucial to managing its progression. X-ray imaging is commonly used to predict osteoarthritis, but manual interpretation can be error-prone due to radiologists' varying expertise. Automated systems using machine learning have been explored for better prediction from X-rays, yet most still lack the accuracy needed for early detection. This paper proposes a more accurate method for early knee osteoarthritis detection using transfer learning models based on sequential convolutional neural networks (CNNs), specifically the Visual Geometry Group 16 (VGG-16) and Residual Neural Network 50 (ResNet-50). Our analysis shows that these models achieve over 90% predictive accuracy, with the VGG-16 model performing the best, reaching a training accuracy of 99% and a testing accuracy of 92%. We also incorporate a Random Forest algorithm to further enhance the predictive capabilities