Machine Learning Models to Predict Kinetic Variables in Cycling
Keywords:
Cycling biomechanics, Machine learning, Modelling, Index of effectiveness, Movement analysisAbstract
This study aimed to predict the index of effectiveness based on lower limb joint kinematics in the sagittal plane and four additional metrics (individual’s mass, power output, pedalling cadence, and horizontal knee position). Seventeen cyclists performed nine submaximal tests of 1 min. Joint kinematics were recorded using a three-dimensional motion capture system and pedalling kinetics were assessed via instrumented pedals. After min-max normalization, several predictor selection methods were applied. The performance of all models was evaluated by 10-cross validation. An artificial neural network model was developed with high accuracy (Adjusted R² = 0.95). Seven multiple linear regression models were developed highlighting a model of 11 predictors (Adjusted R² = 0.86). With this model, the most important predictors that influence the index of effectiveness are known. These models can be integrated into 2D or 3D motion capture systems, which could be useful for bike fitting professionals and trainers to evaluate cyclist's pedalling technique.
Downloads
Downloads
Published
How to Cite
Issue
Section
Copyright (c) 2024 Journal of Science and Cycling

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

