Modeling Power Output for handcycle users in exergaming

This thesis aims to promote fairness and inclusivity, spefically within the context of exercise gaming-a genre of video games, that incorporate physical activity. Zwift, a popular platform for indoor cycling, enables users worldwide to compete virtually. While the platform is optimized for conventional bicycles, it also offers limited support for handcycle users. who are at a significant disadvantage due to the lower inherent power output of handcycling. This often discourages participation and undermines inclusivity. The objective of this research is to develop a model that converts handcycle power into an equivalent bicycle power output, thereby enabling more equitable competition. Data was collected from voluntary, able-bodied participants performing identical tasks on both a bicycle and an handcycle, with fatigue levels standardized using the Borg Rating of Perceived Exertion (RPE) scale. Two modeling approaches were tested. The first involved linear regression with personalized coefficients to estimate equivalent bicycle power from handcycle power. The second approach computer conversion factors for each RPE level based on the ratio of the bicycle and handbike power at the equal fatigue levels. Participants were later tested in real-time Zwift gameplay scenarios using their personalized models. Handcycle power signals were transformed using an Arduino board, and participants repeated the same workouts without seeing real-time power data. The adjusted handcycle power output closely approximated that of the bicycle, improving the fairness of gameplay. Participants reported enhanced enjoyment and a greater sense of equity. Additional tests explored predicting the RPE value from physiological data using linear regression techniques, yielding promising results. Future work will focus on expanding the dataset, identifying factors that influence model sustainability for individuals, and developing generalizable correction factors. These improvements could eventually enable the model's application to a broader range of adaptive athletes, including wheelchair users.