Learning Specific Features for KOA Diagnosis: Autoencoder Network vs. Feature Selection

Knee osteoarthritis (KOA) is a degenerative joint disease that seriously affects patients’ quality of life. Due to the global aging population and obesity, the number of people living with KOA is expected to increase. Although KOA is the dominant disease in Traditional Chinese Medicine (TCM), the clinical diagnosis is currently made by Western medicine based on the patient's medical history, signs and symptoms, and image examinations. This study aims to find specific features for KOA using deep autoencoder and traditional feature selection, thus providing a diagnosis reference and exploring the advantages and disadvantages of both approaches and their interpretability.