Quantifying cricket fast bowling volume, speed and perceived intensity zone using an Apple Watch and machine learning

Area of Science:

• Sports Science
• Biomechanics
• Wearable Technology

Background:

• Accurate measurement of bowling workload is crucial for athlete monitoring and injury prevention in cricket.
• Inertial Measurement Units (IMUs) offer a portable solution for capturing biomechanical data.
• Machine learning models can potentially process IMU data for performance analysis.

Purpose of the Study:

• To evaluate the accuracy of IMUs and machine learning models in measuring bowling volume (BV), ball release speed (BRS), and perceived intensity zones (PIZ).
• To compare the performance of research-grade (SABELSense) and consumer-grade (Apple Watch) IMUs.
• To determine the optimal placement and configuration of IMUs for accurate data capture.

Main Methods:

• Forty-four male pace bowlers participated, wearing research-grade and consumer-grade IMUs on both wrists.
• Participants completed 36 deliveries across two perceived intensity zones (70-85% and 100% effort).
• Gradient boosting machine learning models were employed to analyze IMU data, with radar guns used for BRS validation.

Main Results:

• Gradient boosting models achieved high accuracy for BV (F-score=1.0), BRS (MAE=2.76 km/h), and PIZ (F-score=0.92).
• No significant differences were found between research-grade and consumer-grade IMUs on the same wrist.
• IMUs on the dominant wrist showed improved PIZ classification accuracy.

Conclusions:

• IMUs combined with machine learning provide a valid method for assessing key bowling metrics in pace bowlers.
• Consumer-grade wearables are a viable alternative to research-grade devices for this application.
• Dominant wrist IMU placement is recommended for optimizing perceived intensity zone classification.