Machine Learning Model for Rugby Head Impact Analysis
Author Information
Author(s): Stitt Danyon, Kabaliuk Natalia, Spriggs Nicole, Henley Stefan, Alexander Keith, Draper Nick
Primary Institution: University of Canterbury
Hypothesis
Can a machine-learning model effectively match head impacts recorded via wearable sensors to laboratory-simulated head impacts?
Conclusion
The study successfully developed a machine-learning model that accurately predicts head impact parameters in rugby, highlighting the need for further analysis to align laboratory testing with on-field conditions.
Supporting Evidence
- The machine-learning model achieved high accuracies in predicting impact parameters.
- Most head impacts were classified as occurring to the side or rear of the head.
- Nearly 80% of impacts were similar to laboratory impacts that included neck involvement.
- Field data showed a strong similarity in predicted impact parameters between male and female players.
Takeaway
Researchers created a smart computer program to help understand how rugby players get hit in the head, which can help make the game safer.
Methodology
The study used a machine-learning model trained on laboratory head impact data to predict impact parameters from field data collected via wearable sensors.
Limitations
The study's models do not directly compare time series traces and are limited to the existing library of laboratory impacts, which may not cover all field conditions.
Participant Demographics
Participants included male and female youth rugby players aged 13-17 years from Christchurch, New Zealand.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website