Construction of predictive models for bicycle riding comfort evaluation using electromyogram and electroencephalogram

In the bicycle manufacturing industries, manufacturers attempt to reflect a user's preference, namely riding comfort, on their products. Surface electromyogram (EMG)-based approaches have been researched for evaluation of riding comfort. However, the EMG does not capture user preferences, because it focuses on muscle fatigue, not riding comfort. To solve this problem, we propose an approach that combines an electroencephalogram (EEG) generated from the brain, which controls modulation of feelings and thoughts. Two bicycles that have different parameter settings and two types of tracks (straight and slalom) were selected to determine the riding comfort, especially riding difference, for the first time by using an EMG and EEG. Elastic net logistic regression analysis was used to construct predictive models. The classification accuracy of the bicycles was determined to be 81.9±7.0% for the slalom course. Furthermore, it was demonstrated that the rectus muscle and frontal lobe are important points for evaluation of the riding comfort of bicycles.