Experimental Study Of The Functional Status Of Aortic Valve After Left Ventricular Assist Device Implantation And Development Of A Machine Learning Model

ObjectiveImplantation of a continuous-flow left ventricular assist device(CF-LVAD)results in altered hemodynamics in the aortic root,which affects the functional status of the aortic valve.The primary method for optimizing the functional status of the aortic valve after implantation remains echocardiography-guided pump speed adjustment.However,the optimization method has limitations.Interactions between the heart and the blood pump,as well as methods for out-of-hospital assessment of valve functional status,have rarely been reported.The purpose of this study is to clarify the characteristics of the cardiac cycle in the CF-LVAD,as well as the characteristics of pressure waveforms and flow waveforms in different aortic valve function states;to clarify the effects of physiological indicators and blood pump parameters on aortic valve function states,and to construct a machine learning risk prediction model for AVC states.MethodsIn this study,the CF-LVAD animal model was constructed with six small-tailed sheep,and the heart rate was adjusted and grouped with 50(beats/min)as the initial heart rate,130(beats/min)as the upper heart rate,and 20(beats/min)as the group distance.In each group of heart rate range,the pump speed was changed in increments of 100 rpm,with 2000 rpm as the initial pump speed.Aortic valve functional status,which includes normal aortic valve function(AVO status)and aortic valve closure(AVC status),was assessed by echocardiography.Multi-channel physiologic recorders monitor ECG activity,arterial/venous pressure,and left ventricular pressure;dual-channel ultrasonic flowmeters monitor pump flow.According to the functional status of the aortic valve,the experimental records were divided into the AVO status group and the AVC status group.OriginPro 2021(ver 9.8.0,USA)was used to plot and analyze the waveforms.Wilcoxon rank sum test or two independent samples t-test was used to compare between groups;dichotomous logistic regression was used for multifactorial analysis;machine learning models were built using the R language machine learning package(mlr3);the classification performance and consistency of the models were evaluated by receiver operating characteristic(ROC)curve analysis and Hosmer-Lemeshow goodness-of-fit test;the clinical utility and application value of the models were evaluated by decision curve analysis to calculate the net clinical benefit under different threshold probabilities.Results1.A total of 107 experimental records were generated during the experiment,including 58 cases in the AVO status group and 49 cases in the AVC status group.Observational studies have shown that the CF-LVAD delivers blood from the left ventricle to the aorta continuously throughout the cardiac cycle.Although the flow is not constant,the blood flow is continuous.The hemodynamic changes in isovolumic systole,isovolumic diastole,and filling phase are significant after CF-LVAD implantation compared to the cardiac cycle in the normal physiologic state.2.The cardiac cycle is similar in different states of aortic valve function after CF-LVAD implantation,but there are waveform differences in the ejection and isovolumic diastolic phases.In the AVO status,the aortic pressure peaks twice during the ejection phase and the isovolumic relaxation phase;the pump flow waveform is a continuous waveform with a plateau phase.In the AVC status.the aortic pressure peaks only once during the ejection phase;the flow waveform is a "sharp" continuous single-peak wave.3.Statistical analysis showed that AVC status after CF-LVAD implantation was associated with increased pump rate(OR=1.01.95%CI=1.00-1.01.P=0.002),faster heart rate(OR=1.06,95%CI=1.02-1.10,P=0.003).decreased pulse pressure(OR=0.87,95%CI=0.75-0.99,P=0.049),and decreased left ventricular systolic pressure(OR=0.94,95%CI=0.91-0.96,P<0.001).4.The Ranger model,which included four indicators of left ventricular systolic pressure,heart rate,pulse pressure and pump speed,had an area under the ROC curve of 0.971(95%CI=0.919-0.994)and a Hosmer-Lemeshow goodness of fit test statistic of 4.193,P=0.839>0.05.The decision curve analysis shows that the net benefit of the Ranger model’s predicted outcomes is higher and the decline was smaller when the risk threshold is in the range of 0 to 0.80.Conclusions1.After CF-LVAD implantation,the circulatory blood flow is characterized by continuous blood flow with periodic changes in flow.According to this characteristic,the cardiac cycle can be divided into the fast ejection stage,slow ejection stage,and atrial contraction stage.2.After CF-LVAD implantation,there are significant differences in arterial pressure waveforms and flow waveforms under different aortic valve functional states.3.Significant predictors/influencing factors of aortic valve functional status after CF-LVAD implantation include left ventricular systolic pressure,heart rate,pulse pressure,and pump speed.4.The AVC Status Risk Score after CF-LVAD implantation has good classification efficacy and consistency.It also has some clinical utility.