Comparison of computational cost and prediction error of cardiorespiratory models for exercise simulation

The modeling and simulation of the cardiorespiratory system has allowed a better understanding of the physiological system behavior to different stimuli and pathologies. Among its main applications stands out the computational tools for diagnosis and clinical treatment support, context in which the knowledge of the computational cost is fundamental for its implementation. The purpose of this work is to develop a comparative evaluation of computational cost of the cardiorespiratory models proposed by Cheng et al. and Albanese et al. under physical activity simulation. The evaluation consisted in the comparison of the computational cost and the prediction error of the physiological variables for each model. Variations in carbon dioxide production between 0.2 and 2.0 l/min were implemented, under the conditions and nominal parameters proposed for each model. The variables compared were mean arterial blood pressure, heart rate, frequency of breathing, tidal volume, arterial partial pressure of oxygen and arterial partial pressure of carbon dioxide. The results evidence that the model proposed by Cheng et al. is the most appropriate for the exercise simulation, due to both the best prediction of variables and the best relationship between complexity and simulation time required.