Estimation of pulsatile flow and differential pressure based on multi-layer perceptron using an axial flow blood pump

This study proposes a non-invasive method for estimating the pulsating flow and pressure difference, which uses the blood pump estimation model based on a multi-layer perceptron to calculate the flow and pressure difference under pulsating conditions. The model takes 11 parameters such as the rotational speed, power, and pulsation waveform of the blood pump as the input and uses the pressure difference and flow as the output. The experimental results of 119,590 sample data show that the flow error of the training set of the blood pump estimation model is 0.14 l/min and the pressure difference error is 7.50 mmHg; the flow error of the test set is 0.14 l/min and the pressure difference error is 7.50 mmHg. Compared with the traditional flow and pressure prediction method, this method has higher precision, which will provide a certain technical accumulation for accurately estimating the flow and pressure difference of the blood pump in the pulsating conditions.