Non-Invasive Estimation of Achilles Tendon Load in Fast Bowlers Using IMUs and Machine Learning for Tendinitis Prevention

Fast bowling in cricket is a high-impact athletic activity associated with Achilles Tendinitis due to repetitive loading of the Achilles Tendon (AT) during key motion phases. This study presents an out-of-lab, wearable tendon load monitoring system that combines burst vibration stimulation, inertial measurement units (IMUs), and machine learning to non-invasively estimate AT load during fast bowling. Five bowlers of varying skill levels performed 4 ball deliveries each, with data collected from a custom-built wearable sensor setup in the area around the AT. Signal processing and feature extraction were conducted using MATLAB, data were analyzed, and machine learning models were trained in Python to predict the tendon load proxy. The results demonstrated that professional bowlers consistently exhibited higher and biomechanically more consistent AT loads during Front Foot Contact and Ball Release phases, while recreational bowlers showed erratic patterns. Random Forest models$\left(R^{2}=0.996\right)$and XGBoost models$\left(R^{2}=0.999\right)$yielded high accuracy, with XGBoost better identifying key AT load predictors such as phase duration, phase, and maximum amplitude. The study demonstrates the feasibility of using a low-cost setup of wearable technology to estimate the tendon load outside the lab, providing valuable information to coaches, physiotherapists, and sports science professionals in the management and monitoring of bowler performance.