Artificial intelligence-enabled reconstruction of the right ventricular pressure curve using the peak pressure value: a proof-of-concept study

Abstract

AbstractConventional echocardiographic parameters of right ventricular (RV) function are heavily afterload-dependent. Therefore, the incorporation of the RV pressure curve may enable the formulation of new parameters that reflect intrinsic RV function more accurately. Accordingly, we sought to develop an artificial intelligence-based method that can reconstruct the RV pressure curve based on the peak RV pressure.We invasively acquired RV pressure in 29 heart failure patients before and after the implantation of a left ventricular (LV) assist device. Using these tracings, we trained various machine learning models to reconstruct the RV pressure curve of the entire cardiac cycle solely based on the peak value of the invasively acquired curve. The best-performing model was compared with the performance of two other methods that estimated RV pressure curves based on a reference LV and RV pressure curve, respectively.Among the evaluated algorithms, the multilayer perceptron (MLP) achieved the best performance with an R2of 0.887 [0.834 – 0.941]. The RV and LV reference curve-based methods achieved R2values of 0.879 [0.815 – 0.943] and 0.636 [0.500 – 0.771], respectively. The MLP and the RV reference curve-based estimation showed good agreement with the invasive RV pressure curves (mean bias: -0.38 mmHg and -0.73 mmHg, respectively), whereas the LV reference curve-based estimation exhibited a high mean bias (+3.93 mmHg).The proposed method enables the reconstruction of the RV pressure curve, using only the peak value as input. Thus, it may serve as a fundamental component for developing new echocardiographic tools targeting the afterload-adjusted assessment of RV function.Graphical abstractKey question. Can artificial intelligence be useful in the reconstruction of the right ventricular pressure curve of the entire cardiac cycle using only the peak value as input?Key findings. Multilayer Perceptron predicted instantaneous pressure values with a balanced, low bias throughout the cardiac cycle, even slightly outperforming the reference curve creation methods.Take-home message. Accurate prediction of the right ventricular pressure curve may enable formulation of new echocardiographic parameters targeting the afterload-adjusted assessment of RV function.LV: left ventricular, PVC: pulmonary valve closure, PVO: pulmonary valve opening, RV: right ventricular, TVC: tricuspid valve closure, TVO: tricuspid valve opening