Predicting pedal power on cue

Predicting pedal power on cue

Claudia Doman

18 July 2012: Elite endurance athletes will soon be able to predict what type of training will allow them to reach peak performance at specific times thanks to a new tool developed by a University of Canberra researcher.

Software engineering PhD student Tania Churchill has created a model to accurately predict cycling performance using training and racing data.

A keen cyclist herself, Ms Churchill is looking at the influence that certain training loads and fatigue levels have in an athlete to determine his/her performance.

“Cyclists want to be able to peak at certain times and this model should allow them to design a specific training load plan so they can produce a peak when desired,” she says. “They want to maximise fitness and minimise fatigue and finding out the right balance between these two will help them reach that peak performance.”  

Mentoring program

L-R: UC PhD student Tania Churchill testing her model with cyclist. Photo: Michelle McAulay

Canberra-based professional cyclist Miffy Galloway says this type of information would help athletes like herself achieve maximum performance.

“Athletes and coaches face a constant battle when it comes to trying to ensure athletes achieve maximum performance at major competitions throughout the season,” the 21-year-old cyclist says.

Miss Galloway explains that a specific routine in the lead up to a successful performance in competition early in the year may not lead to the same outcome at another race down the track due to several factors, including: training loads, quality of sleep or diet, among others.

“By using this model, we [athletes and coaches] will be able to know which factors actually contributed to maximum performance on the day and which didn't, allowing us to arrive at the next competition in the same desired physical state than that when we achieved at our best,” Miss Galloway says.  

She says that keeping an accurate training diary is very time consuming and is one aspect of training which is usually neglected by athletes at all levels.

“Having a model which would supplement this process is something that I would be very eager as an athlete to use in my training and have no doubt that it would be beneficial to my coach as well by providing accurate and immediate data when it is required,” she says.

Ms Churchill’s research focused on data from two elite female cyclists who had power monitors attached to their bikes. She followed both athletes’ training and racing data collected by the monitors during 250 and 870 day periods respectively.  

Her results show that when the training data was introduced into her predicting model, the actual performance was closely accurate.

This study measured performance in actual competitions, which is very unusual. “Few studies have used data from real-world training and performance data. This is primarily because of the difficulties involved in capturing and analysing this kind of data,” Ms Churchill explains.

“How do you quantify performance in endurance cycling? Time is irrelevant, and due to the team nature of the sport, the placing of individuals is also often irrelevant. Overcoming these challenges, however, means you have a model which is applicable to the real-world problems athletes and coaches face in planning training programs,” she says.

In the future, her discovery will help coaches develop an optimal training plan by trying different training loads in the model to predict performance and when the athlete will actually peak.

“This model provides objectivity for coaches to make more informed decisions on preparation for their athletes’ performances,” Ms Churchill said.

Furthermore, “it provides memory of how exactly the athlete performed at a certain time, in previous training sessions or competitions, given different factors, which would be invaluable to evaluate what works for that individual athlete,” she added.

The next step is to determine how far out the model can predict performance with reasonable accuracy.

Ms Churchill notes that her model could be applied in other sports, particularly endurance sports such as long distance running, and Nordic skiing. With some adaptations, the model could also be useful for team sports such as AFL or European football. “Coaches of team sports are often interested in gaining an understanding of an individual player’s response to team training”.