WARNING: A Wearable Inertial-Based Sensor Integrated with a Support Vector Machine Algorithm for the Identification...

Area of Science:

• Sports Science
• Biomechanical Engineering
• Artificial Intelligence

Background:

• Subjectivity in race walking officiating can lead to questioned results.
• Artificial intelligence (AI) offers a potential solution for objective fault detection.
• Wearable sensor technology can capture biomechanical data for analysis.

Purpose of the Study:

• To present WARNING, an inertial-based wearable sensor system integrated with AI.
• To automatically identify race-walking faults using machine learning algorithms.
• To evaluate the performance of different machine learning algorithms for fault detection.

Main Methods:

• Ten expert race walkers performed legal and illegal race walking techniques.
• Two WARNING inertial sensors captured 3D linear acceleration data from the shanks.
• Thirteen machine learning algorithms (decision tree, SVM, k-NN) were trained and tested using an inter-athlete procedure.
• Performance was assessed by accuracy, F1 score, G-index, and prediction speed.

Main Results:

• The quadratic support vector machine (SVM) classifier achieved over 90% accuracy.
• The system demonstrated a high prediction speed of 29,000 observations/s when using data from both shanks.
• Performance significantly decreased when analyzing data from only one leg.

Conclusions:

• The WARNING system, utilizing AI and wearable sensors, can reliably detect race-walking faults.
• This technology has strong potential as an objective assistant for referees in race walking.
• WARNING can be valuable for both competition officiating and athlete training.