Abstract
Abstract
Osteoarthritis is one of the most prevalent degenerative joint diseases that causes an estimated 60 million people pain and limited ability to move throughout the United States alone. Due to increasing numbers of people being diagnosed with Osteoarthritis, it is essential to detect the condition earlier than previously possible to provide better patient outcomes as well as lower costs associated with health care. Researchers have been exploring ways to utilize machine learning and deep learning to determine if Osteoarthritis can be identified utilizing a collection of approximately 2350 labeled X-ray images of joints. The researchers utilized several forms of artificial neural networks that included; a stochastic gradient descent optimizer and adaptive moment estimation (ADAM) in conjunction with a variety of machine learning techniques, which included; random forest (RF), logistic regression (LR), gradient boosting (GB), and decision trees (DT). For consistency across all of the models tested, the researcher divided the dataset into a training group (70%) and a testing group (30%). Additionally, the researcher tested each of the models utilizing a wide array of evaluation metrics (auc, accuracy, f measure, precision, sensitivity, and specificity). Results indicated that the ANN-SGD model provided the greatest accuracy (99.50%) compared to the rf model (99.40%), dt model (98.70%) and the remaining models tested. These results demonstrate the capability of both deep learning and machine learning techniques to analyze and process complex medical image data, and further illustrates their potential to transform clinical practices. Providing accurate predictions of osteoarthritis via X-rays allows healthcare practitioners to make informed decisions regarding patient treatment in a timely manner and develop individualized treatment plans for patients with osteoarthritis. Ultimately, this research demonstrates the potential use of deep learning and machine learning in the diagnosis of osteoarthritis and highlights the potential integration of predictive models developed through these technologies in clinical decision making, and ultimately, improved patient outcomes, and represents a significant advancement in the management of osteoarthritis.