Machine learning based automatic diagnosis of rheumatoid arthritis

Abstract

Computer Vision has been an active branch of Artificial Intelligence in the recent years. In particular, gesture recognition is an up and rising discipline that serves to comprehend human gestures. This project focuses on utilizing Machine Learning to perform gesture recognition, specifically fist clenching gesture, to generate automatic risk assessment of developing Rheumatoid Arthritis. To accurately differentiate between hand gestures based on the hand coordinates generated, an Artificial Neural Network is developed to learn weights that map one’s input to the output. This project seeks to research and discuss the possible diagnostic methodologies, and eventually simplify the diagnosis process of Rheumatoid Arthritis by implementing an application which allows users to assess their risks of developing Rheumatoid Arthritis. Results from the trained model produced a high accuracy when recognizing fist clenching gestures. The aim of this project is to implement a more accessible diagnostic method that will help to raise awareness of this illness.