Keywords
Machine Learning, Clustering

Technologies
Python, Machine Learning

Musculoskeletal disorders such as knee osteoarthritis (OA) are a major global health burden, causing pain, reduced mobility, and long-term healthcare costs. Traditional diagnostic methods like X-rays and MRI provide useful insights but are limited by high cost, limited accessibility, and their inability to monitor joint function dynamically.

This project investigates acoustic emission (AE) monitoring as a non-invasive, real-time approach for assessing knee health. AE sensors capture sound waves generated during movement, reflecting cartilage condition, joint lubrication, and mechanical interactions within the joint.

The objective is to determine whether unsupervised clustering techniques can classify knee health based on AE signal features. Time- and frequency-domain features were extracted from signals recorded during controlled knee movements and analysed using K-Means and Principal Component Analysis.

The study evaluates the effectiveness of these methods in grouping acoustic patterns and their potential for non-invasive knee joint assessment.