Predicting Sport Injuries And Player Performance Using Neural Networks

M P Murphy & Associates Pty Ltd

Predicting sport injuries and player performance using neural networks

For predicting risk of injury the network occupies approximately 25 inputs including training loads over a 3-week period, wellness (i.e. fatigue, sleep quality & stress), pain & comfort ratings (i.e. foot, ankle, calf, groin, etc.) and player conditioning. The network consists of the previous 3 season’s worth of data which accounts for anywhere between 6,000 and 10,000 records and the networks are retrained at the end of each season. The results have been exceptional, and have included:

  • Ability to predict soft tissue injuries and player performance during the early in-season games, based on pre-season training programs.
  • Ability to predict soft tissue injuries and player performance on a week-by-week basis during the in-season.
  • Provide the medical and conditioning staff with a tool to search for optimal activity levels to achieve specified outcomes in player performance and injury risk.
  • Assess new player and draft candidates and recommend selections based on required player or team composition.

One of the major sporting clubs recently reported that since the implementation of the neural network model there has been a 57% reduction in player injuries. In one of the most recent seasons the club was also rated #1 in the AFL for the most players to have played every game, least games missed through injury, most players to play 20+ games and least missed games by top 11 players in each team.

After using nD, I wonder how we ever got our jobs done.

The possibilities are limitless.

Check out how our customers are using nD to extract value from data with analytics, then creating amazing applications that deliver business value for their customers.

Tell us how your application is creating business value for you or your customers.

Leave a Reply

Your email address will not be published. Required fields are marked *