A machine learning approach for saddle height classification in cycling

Background: Saddle height is an important factor in bike fitting because it correlates with cycling efficiency and the risk of injuries. Conventional approaches use anthropometric parameters and joint angles as references to calculate the optimal saddle height, such as the greater trochanter height and knee flexion angle. However, these methods fail to consider individual dynamic differences in cycling. Objective: This study proposed a machine learning (ML) model for calculating saddle height based on easily measured kinematic data. Method: In total, 16 subjects participated in riding tests at three saddle heights. The motion capture system recorded the trajectories of markers attached to their lower limbs. Features were calculated using the hip, knee, and ankle joint angles. The optimal feature set was selected using forward sequential feature selection. The accuracies of four ML models were compared using leave-one-subject-out cross-validation. Results: The optimal feature set contained 14 features related to the hip, knee, and ankle joint angles. The sagittal plane knee angle was the most sensitive to the saddle height, with a classification accuracy of 80%. The k-nearest neighbor model had the highest accuracy of 99.79% when using all the optimal features as inputs. Conclusion: The proposed model compensates for the lack of consideration in traditional methods of individual dynamic variations in cycling, providing a more objective tool for data-driven personalization in bike fitting.