Prediction Models of Arterial Pressure during CPR Based on Force Data of Manual Chest Compression

Abstract

Chest compressions are essential for CPR and the quality of it can be evaluated in terms of physical parameters: force, frequency, or physiological parameters: mean arterial pressure (MAP). In this study, an animal model was finished for CPR experiment. The data of the force of manual chest compression and the arterial blood pressure of animals were collected during the experiment, and the force, frequency, MAP and pulse pressure (PP) of manual chest compression were obtained through data processing. In this study, linear fitting was used to solve the linear relationship between force-MAP, frequency-MAP, force-PP and frequency-PP. The relationship between frequency-force-MAP and the relationship of frequency-force-PP was constructed and predicted by the method of RBF and LSTM. As the results showed that the best linear fitting results are frequency-MAP (R-square = 0.58, RMSE = 9.51) and frequency-PP (R-square = 0.62, RMSE = 9.00), and the best prediction results are the PP prediction using the method of RBF (R-square = 0.80, RMSE = 5.73). The method of LSTM got the best result to predict MAP (R-square = 0.74, RMSE = 5.80).