Title:  Analysis of Joint Acoustic Emissions for Health Monitoring with Wearable Technologies

Committee: 

Dr. Inan, Advisor 

Dr. Lee, Chair

Dr. Prahalad

Abstract: The objective of the proposed research is to improve the usability of joint sounds as a non-invasive biomarker for joint health monitoring through machine learning and signal processing techniques. According to the World Health Organization (WHO), musculoskeletal chronic conditions are the leading causes of disability worldwide. Some of these chronic conditions, such as arthritis, have no specific lab tests, and a diagnosis is formed based on a constellation of subjective exams. To evaluate joint health, physicians also listen to joint sounds, scientifically known as joint acoustic emissions (JAEs). As Thomas R. Insel says, ”the good-news stories in medicine are early detection, early intervention”. To enable early detection and health monitoring with wearable devices, we studied JAEs in loaded and unloaded knees during the course of treatment in patients with juvenile idiopathic arthritis (JIA) and developed a classifier to estimate a joint health score. A novel algorithm was developed to detect and exclude rubbing noise and loose microphone artifacts from the signals. It was demonstrated for the first time that JAEs from loaded knees contain different, but clinically relevant, information from those of unloaded knees. By loading joints, JAEs were more sensitive to persistent changes in the micro-structures of articulating surfaces. To further study joint frictions and JAEs generated in different lubrication modes, we compared knee biomechanical signals against synchronously recorded JAEs in a healthy population. The results of this work demonstrated for the first time that joint sounds are highly correlated with knee tribology. The proposed work falls along the lines of assessing knee health scores of short JAE frames that carry the key clinical information. Our goal is to identify phases of the movement cycle that carry the clinically important information. This could potentially benefit the severity assessment in pathologies (e.g. JIA, RA, or OA) as well as monitoring the treatment progress with longitudinal measurements. This deeper knowledge can help optimize the signal processing steps to improve the usefulness of this digital biomarker in joint health assessment.