Functional Data Analysis of the Power-Duration Relationship in Cyclists

Area of Science:

• Exercise Physiology
• Sports Science
• Biostatistics

Background:

• Mean-maximal power (MMP) data in cyclists contains complex variations.
• Understanding these variations is crucial for performance prediction and analysis.

Purpose of the Study:

• To apply functional principal component (FPC) analysis for extracting and modeling conserved variations in cyclist MMP data.
• To develop a predictive model based on identified principal components.

Main Methods:

• Functional principal component (FPC) analysis was performed on a large MMP dataset.
• A 3-parameter model (F3 model) was derived from the first three FPCs.
• Model fit and sensitivity to sprint/endurance bias were validated using out-of-sample data.

Main Results:

• The first three FPCs explained 97% of the variation in MMP data.
• FPCs were identified as gain, sprint-endurance bias, and W' analog functions.
• The F3 model demonstrated excellent out-of-sample goodness of fit, with FPC2 distinguishing sprint vs. endurance bias.

Conclusions:

• FPC analysis is effective for identifying principal modes of variation in MMP data.
• The derived F3 model shows potential for accurate performance prediction in cyclists.
• The study offers insights into the mechanistic factors influencing exercise performance.